Chapter 1 Introduction
1.1 Research Background
Over the past five years, the campus food delivery service sector has seen an unprecedented increase, and fundamentally changed the way students on campus eat and live. The latest statistics show that the annual growth rate of takeaway orders in Chinese universities has been above 25%, and the campus has become a high-density area for consumption logistics. Therefore, the penetration rate of smart food pickup lockers has risen rapidly and now covers over 60% of all the main higher education institutions, which is the primary infrastructure for receiving and distributing these large quantities of food. But this quick building of infrastructure has led to a problem; the sudden increase in the demand for campus food delivery is too much for the current capacity of pickup lockers. Although the hardware has been upgraded, the interaction design and operational efficiency have not improved much, and there is still a problem with the last mile of delivery in closed campus environments.
The above contradiction will be more noticeable due to problems with usability that will have adverse effects on campus users and food delivery workers. For the first group, there are long operating wait times at peak dining hours, and the service experience is poor. Students often have to wait in line at the terminal screen for a long time due to the chaotic nature of the interface and sluggish operation of the system. Furthermore, frequent failures in recognising the pickup code lead to serious problems in the operation of food delivery. A failure of the barcode scanner and a delay in validating the numbers can extend the time to obtain the goods and also influence the temperature and quality of the food, thus failing to meet the basic requirement of the delivery service. At the same time, these problems of usability have similar negative effects on the delivery staff. The courier has a short delivery time, but due to the system delay and the troublesome way to allocate the compartment, they are forced to remain idle at the locker. This reduces the number of items a courier can distribute per hour, increases their work pressure, and often leads to a large number of couriers being delayed at the locker due to traffic jams, thereby affecting the operation time of all parties.
It is necessary to solve these systemic problems in line with the new closed campus food distribution rules. According to the general regulations of campus security and administrative management, most universities have begun to implement closed-campus management rules, and thus, external delivery personnel are prohibited from entering the teaching and residential areas. Under these new rules, intelligent food pick-up lockers will be the only link that can connect the outside delivery network with the inside campus. Therefore, the operation, reliability and usability of these lockers are not only convenient but also help to ensure that campus distribution regulations can be carried out smoothly without disrupting the normal operation of the logistics system. If the locker system is not good enough to use, it is detrimental to the goal of orderly, safe and efficient campus logistics and will lead to more unsanctioned dormitory deliveries by students and couriers in order to avoid the official channels.
Therefore, given the current situation, it is urgent to conduct field-trial-based usability optimisation of campus food pick-up lockers. Laboratory tests and theoretical heuristic analyses often cannot reflect the actual dynamic, large-scale and variable environmental conditions of campus life. Field tests can offer empirical, environmental data that can reveal the operational problems of various user groups at different times and in different circumstances, such as those caused by a large number of people at the peak period, technical failures due to environmental changes and so on. Based on the above, we can establish the specific needs for improvement, such as improving the sensitivity of the interface, simplifying the operation, and increasing the accuracy of code recognition. Evidence-based refinement can help us meet the demand for fast deliveries while adhering to the strict campus management regulations, so that campus food pick-up lockers can become a convenient, efficient and resilient infrastructure.
1.2 Research Objectives and Significance
The main goals of this study are to solve the problems with the usability of Campus Pick Up Lockers from the three levels above. The first is to find the main usability problems of the current campus food pickup locker by means of standardised field tests. Based on the human factors and ergonomics of the university, this study will use a structured observation method, a task-completion measure and a system usability scale to collect empirical data on how people use this system in daily life. At the same time, we will conduct the classification of operations and try to know where problems occur by means of objective indices such as delays in the interface, unreasonable use of space and a lack of clarity in prompting messages to assist users. The second is to propose practical ways of improving usability that are in line with the particular characteristics of campuses. There are different types of campus environment, such as extreme fluctuations in time demands during school breaks, large crowds of pedestrians and specific groups of users' dimensions and perceptions. Therefore, the general design heuristics of commercial lockers are not suitable. Based on the results of field tests, the proposed model will be constructed with all the different factors in scenarios to form an intervention plan that addresses problems in hardware layout, the operation of a user-machine interface and the systematic arrangement of procedures. The third goal is to use a comparative test to show that the above-mentioned improvement plan is reasonable. After some improvements have been made to the plan in the field, we will carry out repeated trials and collect data on changes in the performance of the system before and after the improvement; at the same time, we will also present the data for evaluation by readers.
Theoretically, this paper aims to add to the research on the usability of intelligent pickup devices in different parts of the campus. Current literature on human-computer interaction and industrial engineering is generally about self-service parcel lockers in public spaces or commercial logistics points in the city, and does not focus on the special operation mode of semi-closed institutional ecosystems. Based on the in-depth analysis of various interactions between campus users and food pickup lockers, this paper extends the range of application of existing usability evaluation systems and improves the theoretical basis for high-frequency, time-sensitive autonomous service terminals. Methodologically, the combination of standardised field testing and scenario-limited optimisation forms a reproducible analysis system, thus enriching the theory of contextualised design optimisation and narrowing the gap between the general principles of ergonomics and the constraints of an institution's particular environment.
The research results are also of good practical value and can be used in the day-to-day work of the campus property management team, the locker operation service company and the food delivery platform. For campus property management teams, the refined spatial arrangement and simplified interactive procedures can reduce the crowds of people at peak hours and increase the overall flow of people; they can also reduce the burden of dealing with conflicts on the spot. For locker operation service providers, the resolution of usability pain points will reduce the number of complaints from users, and thus the maintenance costs due to malfunctions can be reduced, and the operating life of hardware assets will be stable. For the food delivery platform, the improvement of locker usability will reduce the time it takes to pick up orders individually. Faster throughput prevents backlogs of secondary orders and avoids the problem of temperature drop due to long stays in the locker, thus reducing the chance of food spoilage. Finally, by adjusting the operation parameters of the smart pickup station to meet the actual needs of campus students, the study can provide practical, industry-standard-based engineering plans to enhance service stability and improve the convenience of food delivery for students.
1.3 Research Scope and Core Definition Clarification
Determine the research scope precisely and clarify the main ideas strictly to ensure that the research methodology is correct and the data is trustworthy. Primarily, the research's spatial and functional scope is limited to the campus self-service food pickup lockers specifically built and placed to store takeaway meals within the university's geographical and administrative service areas. Explicitly exclude general express pickup cabinets outside the school, and do not use the school's smart lockers in areas not for food. Due to the different operation, hygiene, and time requirements of food storage, it cannot be simply stated that the environment is suitable for all general goods logistics. Focus on campus food lockers and study the specific interaction situations between people and machines as well as the environment in the university food delivery system.
To systematically break down the research object, a number of basic concepts need to be clearly defined. First, this paper proposes the idea of "full-scenario usability" for a food pickup locker. In terms of delivery rider delivery and student pick-up, this is not only a simple change in the way goods are distributed but also includes other problems. Full-scenario usability requires the hardware to be easy to use, the software to respond quickly, and the space to be ergonomic; otherwise, they will not be used flexibly in different times of day and under different conditions of the university.
Second, the food-specific division layout in this paper is to divide the storage space according to the needs of takeaway food packaging. Not all express cabinets have the same size; some can be large for big objects, while others can be small for small things, and there will also be some medium-sized spaces, among other features. The dimensions of the various modules should be changed according to the size of different meal containers; for instance, small food containers are designed for single-use lunch, while large food containers are suitable for bulk consumption and can be divided and stored separately. The above arrangement is to make the best use of the room and shorten the waiting time for collection.
Third, according to the behaviour path and function of the user during pick-up, the campus user group is divided. Therefore, the first type is that depositors and retrievers have been divided into two groups; further subdivisions may be made based on the frequency of use and time. Based on the different needs of the riders for depositing multiple items quickly and students for making one-time withdrawals, appropriate solutions should be developed to meet the requirements of usability.
In addition, the empirical boundary of the field test must be determined to ensure that the research is both controllable and practical. Field test boundaries will be based on the operation guide and will not cover extreme cases that are not allowed in campus food delivery. Specifically, when the meal in the locker has been left for more than two hours under high-temperature conditions, it cannot be studied. Such a situation is a serious violation of the campus food safety regulations and institutional hygiene regulations; therefore, it is an abnormal operation failure, not a normal usability problem. By removing these non-compliant outliers, the field test only observes the normal operating conditions to obtain a usable improvement plan that is practical, compliant with institutional rules, and beneficial for daily work efficiency and satisfaction of the campus food pickup infrastructure. This boundary protection will not be distorted by some statistically frequent and policy-prohibited extremities in the empirical data, thus keeping the internal validity of the subsequent analysis model.
1.4 Overall Research Framework and Chapter Arrangement
The main purpose of this paper is to build a research system from four aspects in sequence: proposing problems, doing theoretical work, conducting field experiments and processing the data and coming up with strategies. First, the problem link is to find the practical problems of the current campus food pickup lockers, such as low operation efficiency and poor user experience, so as to build the research direction. The theoretical preparation link forms the basis for the theory of usability engineering and human-computer interaction. The field test implementation link is the empirical part; that is, observe and record actual usage scenarios in the real campus environment. Finally, the data analysis and strategy formulation link will use the empirical results to suggest practical design changes and then verify them in practice to ensure they work well. The four links are closely connected, forming a closed-loop research logic that can be improved from abstraction to practice and back to theory.
According to the above framework, this thesis has been divided into five chapters, each with different research methods and desired results. Chapter One is an introduction to the research background, purpose and significance. Based on the literature review and comparison analysis, this study aims to explain what the research gap is and build a framework. Chapter Two will systematically review the Theoretical Foundation, usability evaluation models and ergonomic standards. Through theoretical derivation, this chapter will produce a set of specific indicators for the usability evaluation of campus food collection lockers. Chapter Three, the core empirical part, is about Field Test Implementation and Data Collection. Through field observation, behavioural coding and the system usability scale, this chapter collects operational data and user feedback; thus, a comprehensive dataset about the actual operation of users and lockers can be obtained. Chapter Four is about data analysis and strategy formulation. Using statistical analysis and root cause analysis, this chapter will find the main problems with usability and give some targeted improvements to the layout and operation of the interface. Chapter Five will be the final part of this paper, summarising the research results, pointing out the shortcomings and suggesting a path for future research, and it is expected that there will be a general conclusion and direction proposals.
The first major innovation of this study is to use the fieldwork-based method, rather than the traditional pure simulation or survey method. While the simulation environment can control the variables, it still cannot simulate the actual unpredictability and cognitive burden of the campus, such as the overcrowding at peak times in the cafeteria and environmental noise. Similarly, the self-report bias of the questionnaire survey is limited; that is, the users' stated intentions may not reflect their real behaviour. The whole-process field test can collect ecologically valid data on the system, so as to understand the actual operation process of the system in daily life, changes in error rates, etc. Based on the above evidence, the problems with the usability of the system and the improvements proposed will be based on the actual circumstances of the system, not on assumptions. Therefore, the full-process field test is expected to strengthen the external validity of the research results and their practical application, and thus to provide a reliable basis for the design of the campus food pickup locker to conform to the requirements of various policies and regulations.
Chapter 2 Literature Review and Theoretical Foundation
2.1 Systematic Review of Usability Research on Smart Pickup Devices
Over the past ten years, as the number of smart pickup devices, especially smart parcel lockers, has increased rapidly, there have been many studies on their usability. According to the retrieval and analysis of domestic and foreign literature, the existing research on usability is mainly divided into three parts: Human-Computer Interaction (HCI) operation performance, spatial layout rationality and scene-adaptive function design. The HCI operation performance module is mainly concerned with the efficiency and cognitive load of the user-device interface interaction. The core idea of this area is to reduce the number of operation steps, make the touch screen respond quickly, and give clear feedback for the completion of the task. Spatial layout rationality is the design of the space in terms of ergonomics and architecture. It is based on the idea that user fatigue and error rates are directly proportional to the physical accessibility of the compartments; therefore, the operating procedures should be based on anthropometric data to achieve optimal visual and reach limits. Scenario-Adaptive Function Design is the core part, and it will consider different times, places and other circumstances to adjust the system; for example, in the case of peak-hour congestion or changes in the size of parcels, the modularised hardware and flexible software structure will continue to provide stable service.
At the same time, the various approaches used to evaluate these dimensions will also have their own strengths and weaknesses. The heuristic evaluation, based on the inspection by experts, can quickly and efficiently conduct some tests in the early design stage. However, its basic problem is that it is based on the intuition of experts, which often differs from the actual cognitive model of users and is therefore unable to address the actual problems of society. Laboratory simulation tests are conducted under strictly controlled conditions to separate specific variables, and thus they provide accurate quantitative data for operation time and error rates. But due to their artificial nature, they lack the ecological validity necessary for understanding the space layout and the ability to adapt to the environment; the simulated environment cannot replicate the psychological stress or physical limitations of a real-world pickup scenario. Field tests are conducted on site to observe how people actually behave under real-world conditions. This way of operation is more suitable for practical use because it can gradually discover potential problems in user experience, such as being unable to be seen in natural light or queuing issues at the busiest time in the laboratory. The main problems of field tests are that they are difficult to implement due to various reasons and have many interfering factors; strong statistical analysis is needed to determine which parts of the two are related.
Based on the above-mentioned three kinds of modules, a general agreement about the common indicators of usability in existing literature has been formed. Operationally, the commonly accepted indicators are Task Time, Error Rate and Cognitive Load Index; together, they can show the difficulties in human-computer interaction. Ergonomically speaking, the reach ability of a compartment, the fit of a visual angle and the flexibility of an interface's height are the three main factors to be judged in the layout of the space. Adaptive mode is used to analyse whether the system can handle the heavy load under various working conditions in normal time, and whether it can meet the requirements in special situations to achieve adaptive function design. These indicators form a comprehensive and multi-level evaluation system, providing a theoretical basis and empirical data for the subsequent field test-based usability research on campus food pickup lockers. Based on the above verification structure, a reasonable and operable path for diagnosis, improvement and design is presented.
表1 Usability Research Comparison
2.2 Current Research Status of Campus Food Distribution and Pickup Scenarios
The study of campus food distribution and pickup scenarios has drawn more and more academic attention because the operational characteristics of campus environments are different from ordinary community settings. At the same time, the campus food pickup site is essentially a narrow space and a short-time logistics node for many similar people. Unlike community distribution, which has a relatively dispersed time and space demand, campus scenarios have an extremely concentrated peak of meal pickup activities. Most of the pickup attempts in the research are concentrated in a narrow time period at noon and in the evening, generally no more than ninety minutes. Because of the severe time clustering, there is a large amount of traffic at once and it cannot be processed by the existing delivery terminals to meet the high requirements.
The campus scenario has relatively low operational proficiency among the users. The user base is a group of students who are familiar with the digital world and can handle automated interfaces well; there are also some other people, such as teachers and campus maintenance staff, who may have problems using touch screens, barcode scanners or the application system. The difference in operation proficiency will affect the time required to complete a single transaction, thus increasing the problem of congestion in a queue during peak hours. At the same time, the literature notes that there are strict requirements for food storage temperature and space independence. Campus food orders have various heat preservation requirements for the dishes; for example, some hot dishes need to be kept warm for a long time, and some chilled drinks need to be refrigerated. Conventional ambient lockers for community use cannot meet the requirements of food preservation and hygiene, and thus cause harm to the food when it is taken out.
Based on the field observations and operational data in the past, many practical problems have been discovered after the implementation of campus food pickup lockers. The most common problem is that there is a lack of space around the lockers at the peak time, which slows down the speed of retrieval and also increases the chaos of pedestrians in the campus. Second, misplacement and wrong retrieval are also frequent occurrences, which are generally caused by the fact that the visual guide and interface prompt are not clear enough, especially for users with relatively low operating ability. Thirdly, due to the lack of active thermal control in the ordinary locker, the quality of the food inside will deteriorate quickly, thereby reducing the users' satisfaction and violating the rules of the catering and food industries.
Based on the above problems in the campus, previous researchers have put forward targeted suggestions. To solve the problem of time delay, researchers have proposed a dynamic capacity allocation model and a staggered pickup schedule algorithm, and have used real-time demand forecasting to distribute the peak load and smooth the demand curve. To fill the gap in proficiency, according to the human factors engineering theory, a simplified interface, some convenient interaction methods such as voice-guided or near-field communication retrieval, and suitable ways to help people remember or recall in the process can be considered to lower cognitive demand and save time. Based on the requirements of thermal and space, the plan is to redesign the locker hardware modularly and add thermally insulated rooms to actively manage the microclimate; a division of space will also be made according to the temperature of the order. Taken together, these works offer a necessary theoretical framework to enhance the convenience of the food pickup lockers in schools and promote user satisfaction by modifying the design to suit local conditions rather than adopting the generalisation of a single plan from other communities.
2.3 Core Theoretical Framework Applied in This Study
The basic ideas in this paper are the User-Centered Design Model, User-centered Design Theory and Scenario-based User Demand Differentiation Model in the ISO 9241 standard for usability evaluation systems. Not only are these theories not abstract, but they also provide an orderly way to conduct the entire research, ranging from the design of field tests to the analysis of problems and finally the development of solutions.
ISO 9241 Standard has presented a basic definition of usability and defined three dimensions for assessing it, namely effectiveness, efficiency and satisfaction. In terms of campus food pickup lockers, the first type is the precision and completeness of picking up food by users; the second type is the amount of resources used in comparison with the degree of precision and completeness; and the last type is how comfortable people are with the system. In the early stage of the field test design, this standard will be used. Based on the three dimensions, the study will define some specific measurement indices and coding systems for on-site observation and later questionnaires. ISO 9241 can be used in the diagnosis of operational problems and cognitive difficulties to classify them according to the standard usability deficiency, avoiding treating them as isolated anecdotes. In formulating improvement measures, this framework will serve as a benchmark to determine whether the proposed ergonomic adjustments have truly enhanced the system's performance rather than transferring the problems from one usability dimension to another.
Based on the above, the research will be carried out according to the procedural orientation of the user-centred interactive interface design theory. According to this theory, the design and continuous optimisation of the system should be based on the cognitive model, behaviour, and physiology of the end user, who is the various campus population here. The main principle is that the design process is a cycle of learning about the use situation, outlining the requirements from users' perspectives, creating a solution plan, and assessing how well the plan meets the requirements. In this study, the theory is used to design field test instruments that can monitor the real-life interactions and cognitive states of students with the locker interface rather than following any fixed mode of operation. In the diagnostic stage, a user-centred theory is used to analyse pain points from the perspective of users' perception and discover where the logic of the system does not meet users' expectations. In the strategy formulation stage, the above will be strictly implemented to reduce cognitive load and give intuitive affordances, so that the improved locker system can better align with the natural behaviour path of users.
Finally, the scenario-based user demand differentiation model is designed to address the dynamic nature of human-computer interaction and different situations. According to this model, the user demands and corresponding usability threshold will vary from one scenario to another. The first is that the students in class are in a hurry and have less time to pick up their food, while the second is that in busy times of large numbers of people in the dormitory, there is no place to put the meal, and the third is that the process of the students collecting their meal after leaving school is relatively simple. Based on this model, we will organise the observation in a stratified way according to different operation situations and not treat them as a homogeneous group. At the time of pain point diagnosis, it can help researchers identify the particular problems in the scene and know that a certain interface is very good in a low-pressure environment but poor in the case of time pressure. Therefore, when developing an improvement strategy, this model is used to select the appropriate one according to the situation, so that some designs can be the most convenient and others have normal functions.
Together, these three theoretical systems form a unified and strict methodological system for this study. The first is that ISO 9241 provides a scientific basis for evaluation, the second is that user-centred design theory places people's needs first in the analysis, and the third is the scenario-based model that considers changes in people's needs under different circumstances. The combined use of the above can make the entire research process, from the design of the field test to the identification of pain points and the formulation of improvement strategies, no longer based on subjective feelings alone, but can also be reasonably improved for the usability of campus food pickup lockers.
表2 Core Theories Overview
2.4 Research Gap Identification in Existing Related Studies
The systematic review of previous studies on the usability of self-service lockers shows that there have been many improvements in general interactive system design and ergonomics. However, when compared with the actual needs of campus food pickup lockers for improvement, there are still some problems in the existing research that have not yet been solved in the application field. There are three basic research deficiencies that have not been able to translate the existing theory into the actual improvement of campus food pickup scenarios.
Most of the previous studies on express pickup lockers have focused on the general characteristics of express pickup lockers, but few have considered the special features of food pickup. General parcel lockers are mainly used to store permanent goods, so time urgency and environmental control are not very important. Food pickup lockers are in a dynamic service environment with strict time limits, temperature requirements, and high hygiene standards. Usability of the food locker is closely related to the preservation performance and urgency of pickup; if it is too late or has been damaged by heat, it will be considered a serious usability failure, and this aspect has not been addressed in the traditional design of locker usability. Therefore, applying general express locker evaluation standards to the food pickup scenario will result in an underrepresentation of the structure and fail to include the important time and quality aspects that users care about in food service.
Most of the usability testing processes at present are not based on the data of all the actual tests conducted in the field. Current research is mostly focused on laboratory-type simulation experiments, heuristic evaluation and simulation task analysis. Although these methods can guarantee the internal validity of the experiment and control some variables, they will inevitably reduce the external validity of the human-computer interaction in an unregulated social-technological environment on campus. The noise in the environment, crowding stress, network delay and peak-hour distribution of the load will all have different effects on people's behaviour and attention in daily life. Without the actual operating constraints and changes in people's behaviour in all kinds of circumstances of daily life, the current assessment system will be unable to present all the problems that arise in life, so the suggestions for improving the system may not be able to address these practical problems well.
Thirdly, there are few studies on the differentiated needs of different types of campus users and how to address them. The operational characteristics and usability requirements of campus students vary greatly. The goal of students living on campus is to reach school as quickly as possible in the early morning rush hour to avoid being late for classes; faculty members and other off-campus commuters want a clear, error-tolerant and transparent system in the middle of the day. The majority of the literature employs a single-user model to study the differences among user demographics and work situations. This will hide the usability problems of subgroups, and thus a targeted and differentiated improvement plan that meets the different ergonomic and cognitive needs of different campus groups cannot be developed.
To address the above deficiencies, the research will be systematically extended in the research direction to make up for the shortcomings. Redefine the evaluative structure to include food-specific attributes, apply strict full-process field testing methods to collect real data on actual operation, and carry out stratified differential demand analysis among different user groups, thus providing a context-aware, empirically based optimisation system for campus food pickup lockers. In this way, the gap in theory can be filled, and the practice of human-machine interaction in the campus service facilities can be improved.
Chapter 3 Design and Implementation of Field Tests for Campus Food Pickup Lockers
3.1 Selection of Test Sites and Sample Participant Groups
The selection of the test area and the population of the sample is the basis of the field test, which will directly affect the ecological validity and reliability of the empirical data. For the food pickup lockers on campus, there are different distributions of space, time, and people in various areas. Therefore, the selection of test locations should follow the principles of representativeness, stratification, and diversity of operations to understand the interaction of people with lockers under various environmental conditions.
According to these principles, three typical campus food pickup locker sites in a comprehensive university were selected. The first place to be tested is next to the student dormitory. Geographically, the area is located at the intersection of densely populated residential areas; there are many pedestrian paths and little buffer space. There is a significant change in the number of people using it during the day, and the peak time is around noon and dinner time; during this period, there are usually thousands of people. At present, there is some congestion during peak hours, and the frequent timeout problems have caused people to crowd the lockers; there is also a lack of space around the queue. The second place to be tested is near the campus teaching building. A temporary, schedule-based movement pattern is presented in this place, and the users are moving at high speed and not staying long. There is no steady flow of users throughout the day; it is irregular and has many small peaks scattered in time with the school year. The turnover rate of the lockers is high due to the daily demand for school, but sometimes there will be sudden changes in the number of people, and thus, problems will occur. The third place to be tested is in the faculty residential area of the campus. The geographical environment here is scattered and peaceful. There is no peak in the flow of people; it is relatively small and distributed evenly throughout the day. At present, the usage rate is moderate, but there are still some obstacles in use for older teachers; for example, the interface is not easy to read, and the operation takes too long.
To ensure the statistical power and representativeness of the field test, the sampling standard for participant groups was decided through stratified random sampling. The number of valid participants is 120, which are all divided into four groups of users, namely, resident undergraduates, postgraduate students, on-campus administrators, and food delivery workers. This stratification can reflect the different behaviour habits and work needs of each group. Resident undergraduate students account for the largest number of people; although they can easily use the Internet, they are not satisfied with slow work. The schedule of Postgraduate students' collection has been disrupted due to research; On-campus administrative faculty are used to working during the day but lack computer technology; Food delivery personnel work on the supply side and focus more on the convenience of the locker than the appearance of the locker.
The 120 sample participants' demographic and behavioural characteristics have been strictly controlled. The age range of the sample is from 18 to 60 years old, and it can be divided into four age groups: 18-22, 23-30, 31-45, and 46-60, to reflect the proportions of all age groups among digital natives and older people who may have difficulties in human-machine interactions. The daily frequency of food pickup service usage will be divided into light use, moderate use and heavy use, and the frequency threshold will be set at less than once a day, one to two times a day, and more than two times a day respectively. The difference in operation speed can be used to test the learning curve of beginners and the operating speed of habitual users. In addition, the degree of operation of the device was tested to determine whether it was basic, intermediate or advanced. Basic proficiency refers to users who can only retrieve simple code; Intermediate is those who are proficient at navigating through many interfaces; Advanced is users who are good at solving problems and have mastered the auxiliary functions. Based on the above, the field test can cover all kinds of users and thus is reasonably representative for a follow-up usability improvement study.
3.2 Construction of Multi-dimensional Test Index System for Usability Measurement
Multi-dimensional test indicators can better reflect the problems with the campus food delivery lockers so that they can be improved systematically. Based on ergonomics and human-machine system engineering, an objective, quantitative index system will be constructed to describe the relationship among users, hardware interfaces, and space. The system has four first-level indexes, namely, human-computer interaction performance, spatial layout adaptability, scenario function suitability, and overall user satisfaction. The multidimensional structure can show all the problems of the field test in terms of objective operation efficiency and subjective psychological response.
Human-computer interaction performance will be decomposed into 5 secondary indexes for the purpose of evaluating the dynamic process of interaction. The first index is single operation time, and the time required for the screen to be turned on and open the locker door is about 10 seconds; that is to say, it should be the fastest. The second is pickup code recognition success rate; if the number of successful code authentications is divided by the total number of attempts, and the standard is above 98%, it can be inferred that the sensor and algorithm are reliable. The third is the misoperation rate, which is the proportion of invalid touch inputs among all screen interactions; according to the standard, it must be less than 2%. The fourth is the interface information acquisition time, the time when the eyes are on the primary input prompt and stop; the acceptable time should be less than 3 seconds. The fifth is the time to recover from an operation error, which is the time needed to correct an incorrect input and re-enter a valid code; it should be less than 5 seconds, reflecting the fault-tolerance of the system.
Spatial layout adaptability is to examine whether the physical locker facilities and the size characteristics of the campus people are appropriate. It is divided into three sub-indices. First of all, the coverage rate of the visual reach area of the screen refers to the proportion of the screen that is in the optimal binocular vision range; that is to say, more than 90% of the area meets the requirements. The second is an operational reachability index; that is to say, whether the manual control and locker handles fall within the reachable workspace envelope or not; based on the joint torque model, we have set the requirement that more than 95% should be accessible without postural strain. Third, the queuing space efficiency is to measure how many square metres each person can use at the time when there is a large number of pickups, and the evaluation criteria are that each person should have at least 0.8 square metres of space to avoid affecting operations.
The Scenario Function Suitability is to verify how well the system's functions work in various situations on campus. It has three sub-indices. First, the success rate of abnormal state processing is to see whether the system can solve problems such as expired codes and network timeouts; the resolution rate should be above 95%. The second is the ability to handle concurrent requests; that is, how many pickup requests can be processed simultaneously by the server at the same time, and a server response time that is within the acceptable range of no delay for 50 users is required. Thirdly, the resilience to environmental interference is to test whether the operation will be maintained normally under various conditions, such as changes in outdoor light, and then determine whether it is successful.
Overall user satisfaction is a sense of psychology and a measure of cognitive burden in an activity, and it can be divided into three sub-indicators. The first is that the System Usability Scale has been modified to collect answers from people in a regular survey, and the threshold is 68; scores above 80 are considered high usability. Second, the cognitive load index is used to assess perceived mental effort in the pickup task and has been normalised by NASA-TLX to a maximum acceptable value of 50 (out of 100). Thirdly, the willingness to re-use rate is a long-term evaluation of the users' feelings; if the proportion of users who have indicated a strong will to re-use exceeds 85%, it will be considered satisfactory. By strictly defining the measurement methods and standards for each quantitative index, this system provides a solid methodological foundation for identifying usability problems and gradually improving the design of campus food pickup lockers.
表3 Campus Locker Usability Test Indices
3.3 Deployment of Field Test Processes and Data Collection Methods
图 1 测试阶段与数据流采集
The field test process for campus food pickup lockers has three stages in order and is divided into pre-test user information registration, formal guided operation test, and post-test experience interview. The first part will record the user information of the people, the second part will test their behaviour, and the third part will evaluate the usability of the locker by means of their cognition. The first step of pre-test user information registration is to gather the basic data and behaviour of all the people. Participants are required to provide basic personal information and past experience with similar self-service facilities, so that we can later conduct a stratified analysis of usability barriers among various user groups. The second stage is an empirical test of the formal guided operation, which is the core of the field evaluation. Participants will be instructed to retrieve a simulated food order from the locker in a standardised manner and, at the same time, experience the hardware interface and system logic naturally. In order to reduce the Hawthorne effect and maintain the ecological validity of the research, the operational guidance is limited to the initial instruction of the task, and user-system interactions can develop independently. The last stage is a post-test experience interview to discuss what the reasons are for the changes in operation and collect users' opinions.
Along with the construction of the test structure, there are also many ways to obtain data; in short, the above methods are to check if the usability test is reliable. The first data stream is a high-definition camera that records the whole operation process of the user. The visual record will show large-space movement and small-space action at the interface and will not affect normal operation, thus ensuring the authenticity of the behaviour. The second data stream is that the background system automatically collects operation behaviour data. The automatic telemetry has been used to collect the event-level metrics such as the time of the transaction, the duration of the step, the error rate and the recovery path, etc., which can be considered as an objective and detailed digital footprint of the interaction between people and systems. The third data stream is to distribute an on-site questionnaire to collect user subjective satisfaction scores immediately after finishing the task. Based on the existing psychometric scales, this tool can be used to measure the perceived efficiency, cognitive load, and intuitiveness of the interface, so as to better connect objective performance data with people's feelings. Finally, semi-structured interviews are conducted to gather individualised feedback from users. Through this qualitative approach, one can discover new issues, particular problems and constructive improvement suggestions that are difficult to find through rigid quantitative methods.
Therefore, according to the relevant regulations on campus public spaces and the requirements for the protection of users' personal information, it is impossible to conduct a field test for these processes or collect multiple sources of data. The installation of observation equipment, particularly high-definition cameras, is in line with the rules of space management for public monitoring, and the recording area has been clearly marked around the locker interaction area; there are signs at the entrance to show where it is. In addition, all the ways of collecting, storing and using personal information must comply with the rules and regulations of the institution's data management for the protection of personal information. Anonymize the data of the participants and obtain explicit consent from them before registering them in the pre-test. The design of the experiment is in line with these regulations and ethics, so it can provide strong empirical evidence for improving the usability of campus food pick-up lockers and meet the ethical and institutional compliance requirements for human-subject research in the campus environment.
3.4 Pre-test Calibration and Control of Confounding Variables
Pre-test calibration and the strict control of the interfering factors are the basic means of empirical field tests to determine whether the observed changes in usability indexes are due to the design of the campus food pickup lockers or other reasons, such as the system itself and the environment. According to industrial engineering and human factors, pre-test calibration is the adjustment of all kinds of hardware and software to the best state before formal data collection. The above steps are mandatory, and if there is any change in the function of the equipment-such as a hidden defect in the hardware-that will increase the variability and lower the reliability and validity of the test results.
The work of pre-test calibration is conducted in three main ways. First of all, the sensitivity of the touch screen modules should be carefully tested. Touch screens with latent responsiveness or localised dead zones will artificially increase the time taken to complete tasks and lead to errors that do not represent the actual use of the device. Calibration protocols should be uniformly applied for all test units to ensure that the touch sensitivity threshold is consistent in response to different levels of pressure and contact areas. The second is to improve the recognition accuracy of the pickup code scanning module. The scanning module is the first input gateway of the pickup authentication process; hence, it is necessary to optimise and standardise its decoding speed and resistance to angular distortion, ambient light interference, and code degradation. Adjusting this module can ensure that all devices have the same optical recognition capability and avoid the problem of some hardware-specific scanning failures distorting the usability data. Thirdly, all compartments of the lockers need to be empty initially to maintain consistency before the start of each test. This will minimise the effects of different door tension and other mechanical factors due to varying numbers of compartments in the initial state, so that all people involved can start from the same place in physical and systemic conditions.
In addition to hardware calibration, control of confounding variables needs to be carried out to isolate the effect of the locker's interface design on user performance. If not controlled, the confounding variable will distort the true causal relationship between system design and usability indicators. Specific control measures will be operated in accordance with the following steps. First, the temperature of the test environment should be kept constant at 20 to 26 degrees Celsius. According to the literature on environmental ergonomics, extreme temperatures can impair people's motor skills and cognitive functions, so we need to adjust the temperature to prevent negative physical effects from reducing the rate of participation in the experiment. Second, choose a time with good weather to carry out the test and avoid heavy rain and strong winds; otherwise, participants will have difficulty reaching the lockers and see the screen clearly, or be affected by psychological reasons, thus reducing the number of pickups. Third, in order to reduce the possibility of people's brains being affected by various reasons and distractions while interacting with the interface, all non-functional information on the interface, such as system updates or advertisements, has been removed. The interface shown to the participant is only related to the functional task flow and does not include other extraneous elements that may influence the cognitive load measurement. Finally, the operation guide given to the different participants by the test staff should be the same. It is done by the standardized Scripted Instruction. Verbal tone, speed, and content are strictly regulated to avoid the experimenter effect; thus, no participant will be accidentally given an advantage or more detailed guidance. By strictly implementing the above calibration and control measures, a sound methodological basis is provided for the subsequent improvement of usability in the field test, and relatively reliable and valid empirical data can be obtained.
表4 Pre-test Variable Calibration
Chapter 4 Analysis of Field Test Data and Diagnosis of Key Usability Pain Points
4.1 Descriptive Statistical Analysis of User Operation Behavior Data
Descriptive statistical analysis is the first method of quantitative analysis for usability engineering; it will organize and describe the data collected from field observations to know the central tendency and dispersion of some user-operation parameters. In the study, it will be an essential analytical tool to convert the original, unorganized data about the operation of campus food pickup lockers into objective and measurable usability indicators. A strict application of this analysis procedure to the 120 groups of valid user operation behavior data obtained from the field test has been carried out, and the basic performance profile of the pickup system has been determined, thereby finding some problems in operation that are contrary to the usual ergonomic and usability standards.
The first quantitative usability index that can be used to measure the time it takes a user to complete the entire pickup process is the average time spent. The above two are time-based indicators; they reflect the level of operational smoothness and the mental load caused by the man-machine interface. The mean completion time, standard deviation and percentiles of the data are shown in a chart, and it can be seen that the data are skewed to the right. A large number of people were late in their operations, and only a small number of people could complete the tasks quickly. When the empirical average is compared with the expected usability level of self-service terminals, a systemic error in the way they interact is identified in numerical terms.
At the same time, the statistical value of the misoperation rate for various operation links also provides a detailed way to determine interface design problems. Descriptive statistics will be used to obtain the frequency and proportion of errors in different operation stages, such as incorrect screen touches or wrong sequences. The chart of the rate change in the entire operation process can be used to find that there are several points of cognitive and motor errors in the interaction process with users. In addition, the failure rate of pickup code recognition is an important technical usability index. The proportion of unsuccessful first-time scanning and manual entry in all the trials is calculated to analyse the perceptual and algorithmic problems of the code-recognition module. The failure rate is statistically significantly higher than the industry standard and has formed a large "friction" point that will not be convenient for the normal use of the lockers.
Spatial usage efficiency can be measured by how much various sizes of locker compartments are utilised. Descriptive statistics show how often different sizes of small, medium and large compartments are used, and a distribution diagram will be provided to show the differences in distribution. Therefore, this index is particularly important for checking whether the current distribution of compartments is in line with the actual volume of campus food deliveries. An obvious difference is that the utilisation rate of large compartments is too high, but small compartments remain idle for a long time; thus, there is a structural problem in the physical design of the system that has led to lower throughput.
Finally, the overall score of user satisfaction obtained from post-task questionnaires is subjected to descriptive statistical analysis to obtain the subjective feelings that may be caused by objective indicators of operation. The average, median and mode of the user's overall experience and the variance of these values will also be displayed. Based on the objective time and error indexes, the distribution of subjective satisfaction and the objective time and error indexes are compared, and it is found that the correlation is significant in terms of operational friction and user satisfaction.
Based on the above descriptive statistics, some usability requirements have been identified and associated; the reasons for the low performance will be analysed. According to the above, the operation links with high statistical misoperation rate, high pickup code recognition failure rate, and average completion time far above the expectation will be the first choice for addressing the usability problem. Based on the above observations, we can provide an objective basis for improving the ergonomics of ergonomics and the functions of the system.
4.2 Correlation Analysis of User Satisfaction and Operational Efficiency Indicators
In order to systematically evaluate the operational state of campus food pickup lockers, it is necessary to understand the connection between objective operational efficiency and subjective user feelings. Correlation analysis is a basic statistical method that can be used to determine how changes in various operating indices are distributed in the general distribution of user satisfaction. Therefore, this analytical path is needed to take further steps from the first one of describing the problem, to make more thorough and reliable investigations into the problems. Determine the degree and strength of statistical significance for this relationship; then, find which part of operation is most unfavourable for users' experience and optimise it through engineering.
The Pearson correlation coefficient was used to analyse the linear relationship between the overall satisfaction scores of users, obtained from verified psychometric scales, and a number of continuous operational efficiency indicators collected in the field test. These indicators are the single operation time, the failure rate of pickup code recognition, the System Response Time and the time between locker doors reopening. Pearson correlation coefficients range from -1 to 1 in absolute terms; their absolute value indicates the closeness of the relationship; and the sign of the coefficient is positive or negative. Then, a two-tailed t-test was used to verify the statistical significance of each correlation, with an alpha level of 0.05 to strictly control for Type I error. Only correlations that had a p-value less than that are considered to be statistically significant; therefore, it is believed that the relationship observed in the data is due to true changes in operation and not random errors caused by sampling in the field test.
Empirical results show that different operating parameters will affect the overall user experience. There is a strong, statistically significant negative relationship between a single operation time and user satisfaction, and it has been found that long operation times can be a source of user dissatisfaction. It can be concluded that cognitive load and perceived time waste are inversely related to the ease of use of the system. Similarly, there is a strong negative correlation between pickup code recognition failure rate and satisfaction. Recognitive failure needs to be addressed; thus, manual code entry or seeking help will interrupt the operation and cause problems in use. The system response time also showed a significant negative relationship, with a moderate correlation coefficient; however, although delays reduce the experience, their immediate impact is less severe than that of active task failures.
The statistics test was used to remove the operational indicators with no significant connection to the overall usability. The re-opening interval of the locker door is an operational index that shows the length of time required before it automatically locks again after an item is taken out; thus, there is no significant relationship. Therefore, according to the current operating parameters of the campus lockers, there is little change in the door interval, and thus the user's overall perception of the system has not been much affected. By systematically filtering out such statistically insignificant variables, the diagnostic process will not waste engineering resources and avoid misdirecting the improvement work to factors that are empirically unrelated to the user experience.
Finally, the above correlation analysis has clearly shown the reasons for the changes in usability in terms of operational hierarchy. It can be confirmed that the time of a single operation and the recognition failure rate of pickup codes are the main reasons for the decline in user satisfaction, and isolated mechanical timings have little impact. In terms of statistics, this is necessary for the industrial engineering optimisation of the system; otherwise, we will prioritize improvements to the algorithm and interface to shorten operation time and increase recognition reliability rather than marginal changes to mechanical thresholds. Quantitatively, it can be more accurately analysed which parts have poor usability problems through this analysis and provide a basis for the next iteration.
表5 Satisfaction and Efficiency Correlation
4.3 Classification and Induction of Typical Usability Failure Scenarios
In the context of industrial engineering and human-machine interaction design, a usability failure scenario is a particular action sequence that causes task interruption, efficiency reduction, or target deviation in the process of user-system interaction. Based on the above, by analysing all parts of the operation videos and user feedback records collected in the field test, the typical usability failure scenarios of campus food pickup lockers can be divided and inferred. The above cases can help map out how users learn to use the system interface and how to prioritise later changes.
The first is misoperation due to unclear direction on the interface. The field test has shown that users often feel confused about the sequential logic displayed on the touch screen, particularly when they need to enter pickup codes or select payment methods. The occurrence process generally starts when there are no obvious signs or clear instructions on the screen for the next step required by users; then, they may enter data in an invalid area or prematurely close the transaction. It has been found that this increases the time to complete a task and makes it more difficult to concentrate. According to the statistical data, this situation has occurred 47 times, accounting for the largest proportion among the observed failures.
The second category is that the food box is squeezed and damaged due to unreasonable partition size. The user is trying to put food in a box that does not meet the volume or shape requirements for the space. Due to the fixed physical limits of the locker grid and the lack of a changeable size feature, people have to push the door shut by force; consequently, the items within may be damaged. The negative effect is that damaged goods can cause direct economic losses, and when the package is broken, the door sensor fails to detect it, resulting in false reports of available spaces. Thirty-one cases of this situation have been recorded during the test.
The third is the operational congestion that occurs at the time of peak pickup. It is a problem that arises due to a mismatch between the arrival rate of users in a certain place and the processing ability of the system. When there are many users gathered at the locker station, and due to reasons such as a slow interface or not enough space for people to walk around, the stay time of each user becomes longer. The subsequent users are not allowed to access the scanner or the screen due to the above reasons. It is not linear; therefore, if we do not improve this problem soon, the overall efficiency of the system will decrease and users will be unhappy because they have to wait too long. There were 28 cases of this kind of situation in the test cycle recorded by the observed data.
The fourth case is that after opening the wrong cabinet door, the user cannot find their own compartment quickly. It is mainly because of the lack of spatial orientation caused by system delays. At first, a user mistakenly opened an adjacent compartment, possibly due to being close to it after scanning, and then saw a number of visually similar doors without any clear spatial indicators. It will take a long time for the user to look at all the doors and find the correct one with light. The negative effects are that it will take longer for one to pick up items in rush hour and there will be more people waiting; that is to say, there will be more congestion. In the field test, this case was observed 22 times.
In short, the four typical usability failure cases mentioned above have been observed 128 times, so a strict basis for experience has been set up to diagnose the underlying reasons for the construction and cognition problems of the food picking-up locker in the campus. The systematic division can also show the limits of the problems in current running and give relevant direction for improvement in ergonomics and structure.
表6 Typical Usability Failure Scenarios
4.4 Divergence Analysis of Usability Demands Across Different Campus User Groups
4.4.1 Comparative Analysis of Usage Demands Between Resident Students and On-campus Faculty
In terms of usability, there are different needs between resident students and on-campus faculty; thus, these two have been used to conduct field tests for data analysis. Resident students often use campus food pickup lockers more frequently every day and frequently come to the system many times in one day for different meal deliveries. On-campus faculty tend to go there less often, only during breaks for lunch. Different times of the day are expected based on different behaviours. According to field test data, resident students are relatively tolerant of long queues and can wait for pickup for about five minutes on average; in contrast, faculty are limited by fixed class times and short breaks and thus have a lower acceptable waiting time limit, usually at two minutes.
Resident students have a strong and urgent need for the function of keeping food warm. Their daily life is fragmented and irregular; therefore, after being delivered, the food is left in the locker for a long time before it can be collected. Thus, keeping it warm is very important for the usability of residents. In contrast, on-campus faculty members' demand for food thermal preservation is generally small because their daily schedules are fixed; thus, they usually pick up meals immediately after delivery, and the time difference between delivery and collection is small.
The preference for the first type of operation is not the same. The majority of the resident students are digital natives, so they want a simple, quick way to operate. These are all very few-operate ways to obtain information, so transaction speed is relatively faster; thus, simple operation is preferred. Although on-campus faculty members are good at technology, they prefer simple ways to use the interface. A visual prompt will be given for each stage of the transaction, and a step-by-step view will also be provided; therefore, they will feel that the operation is too long and troublesome.
Different operation error tolerance levels also exist. Resident students have a relatively high tolerance for operation errors, and they will adapt quickly to any problems that arise in the interface due to frequent learning among peers. Faculty members, due to professional time constraints and poor knowledge of the particular system architecture, are much less tolerant of operating errors. Systemic uncertainties or poor user experience will make me abandon the task quickly.
The reason for the difference in demand is the different circumstances and daily life schedules of each group. The environment is open and flexible for students, and they need to move around the campus frequently for meals, etc. It is used frequently and in an irregular manner, so it needs to be speeded up and be more environmentally friendly. Faculty members have a relatively fixed working schedule, so they are in a relatively high-pressure working environment, and locker interaction is just one part of their work, and they also have to deal with other urgent matters. It is used in such a way that there is no transition time between different tasks in teaching.
In light of the above observations, it can be concluded that the requirements for differentiated usability optimization have become clear. Resident students generally need high-throughput processing speed to handle many search requests at once, strong and stable thermal management systems to maintain the quality of food for a long time, and smooth and fast interface operations. The school's teachers want the system to have clear and simple feedback, be as little as possible affected by the schedule, and have strong systemic reliability to prevent interaction errors. Therefore, to build a universal, reliable campus food pickup facility, we need to consider these different needs and provide solutions systematically.
4.4.2 Difference Test of Operation Preference Between New Users and Frequent Users
In the field test of the pickup locker at university canteen, there are obvious differences in cognitive load and operating habits among different groups of users due to their varied levels of contact with the system. In order to measure these variations systematically, two different groups of people were selected as samples: new users (less than five times to use the campus food collection lockers) and frequent users (more than thirty times to use the same locker). According to the operation preference data obtained from field observation and post-task questionnaire, a difference significance test will be carried out; that is to say, it is an independent sample t-test. We analysed four major indicators: the choice of operating path, the degree of reliance on voice guidance, how sensitive it is to the size of interface text, and whether it can handle abnormal operation problems.
Based on the above analysis, there is a statistically significant difference in the path choice between the two groups (p < 0.01). New users are mainly linear and sequential in navigation, and they will first read all the instructions on the top of the interface before entering the pickup code. The frequent users have an optimised and non-linear path, directly targeting the input box and using keyboard shortcuts or old input data to skip the middle part. Therefore, new users need a very clear visual hierarchy because they are not used to this way of thinking; frequent users are fond of simple interfaces to speed up their work.
There is also a significant difference in how much we rely on voice guidance (p < 0.05). New users need strong multi-modal feedback; that is to say, they want to hear a voice message to know that the locker has received the code and is about to open, and this way can reduce errors at the beginning of operation. Frequent users have grown tired of hearing the voice of the guide, and in places such as the library that are relatively quiet in the university, it can be distracting and annoying; hence, they would prefer to have independent visual confirmation. This shows that there is a certain conflict in the design of auditory feedback: it should be prominent enough to guide novices in operation without bothering experienced users.
Interface text size was sensitive in the first part. New users have mentioned that large fonts and high-contrast icons are needed to reduce the time spent looking for information and the number of errors when entering data, especially in the outdoor glare environment. Frequent users with automated perceptual-motor schemas are not very sensitive to a reduction in the size of text that is still legible; however, they have a strong desire for a compact layout, which reduces the distance of visual scanning and can show more operation information at the same time.
The ability to deal with problems in normal operation, such as wrong codes or stuck locker doors, is different. New users often have problems with cognition during anomalies and often need to be helped by others or just give up. Based on their preferred data, we need to incorporate an explicit, step-by-step diagnostic troubleshooting guide into the interface. Frequent users have strong heuristics to solve problems, such as re-scanning the pickup QR code or manually triggering the retrieval protocol, and they prefer streamlined error handling options without too much detail in the explanatory prompts.
From the results of the difference significance test, the targeted usability design requirements for each group can be obtained. The new user group is likely to make mistakes due to improper operation; therefore, the main requirements for the design are to prevent errors, provide cognitive scaffolding and guide them to solve problems. Therefore, it is necessary to set up an obvious instruction for it and increase the volume of visual and auditory feedback as well as error-recovery paths that are easy to use. The frequent user group is to increase the speed of operation; thus, the main optimisation needs of the user's personalisation are to cut down on the amount of interaction and optimise them more reasonably. Therefore, it is necessary to build shortcuts to get there quickly, keep the information display brief, remove unnecessary modal alerts, and provide a rapid manual override for problems. To meet the above requirements and improve the overall usability of campus food pickup lockers, an adaptive interface strategy that can dynamically adjust system feedback and information density according to the user's interaction level should be employed.
Chapter 5 Targeted Usability Improvement Strategies and Effect Verification
5.1 Modification Strategies for Human-Computer Interaction Interface of Pickup Lockers
The user and campus food locker have the Human-computer Interaction Interface as their communication channel, so the speed and accuracy of taking food should be improved. Based on the usability pain points of the field test, the target modification plan for the interaction interface needs to be established in a planned way to reduce the cognitive burden and operating friction of the system, and also deal with the changes in environmental factors in the open-air campus environment. The above strategies are all based on the basic idea of cognitive ergonomics and human factors engineering to make the interface logic conform to the expected model of the user group.
The first change is to reduce the number of pieces of information shown on the interface. According to the results of the field test, people often feel that there are too many options and find it difficult to choose. Therefore, the design of the interface should follow the usage frequency, and the "pick up with a code" function should be placed in the most noticeable position of the main interface. By raising this main task node and systematically de-prioritising the ancillary functions, the visual scan path is greatly shortened and will accelerate the initial direction selection phase of the interaction sequence.
Moreover, in the event that the external light is constantly changing and thus makes the identification of text easier, there should also be some changes to the size and colour of the font. The interface must meet the existing accessibility standard, and all the typographic elements should have good visibility in both the bright and dark parts of the environment. This requires a dynamic contrast ratio that is greater than the minimum required for reading outdoors and can be scaled up or down according to different viewing distances. Therefore, the above calibration can reduce the problem of missing information and alleviate the visual fatigue caused by prolonged concentration in the environment at noon.
In addition to the change in perception, simplifying the operation syntax will also improve the fluency of interaction. Remove the redundant middle operation steps to reduce the number of motor movements and thus lower the perceived system latency. Grouping all the confirm dialogues of the layers into a single, successive thread sequence can make the path of interaction more certain and thus reduce the occurrence of sudden task interruptions, thereby speeding up the total throughput of the pickup system.
In addition, after a person has been authenticated in the physical locker matrix, the space orientation of that person is a cognitive problem. The dynamic animation guide will be shown immediately after the code is correctly entered, so that it can be understood intuitively. Do not just use a static number and peripheral lighting; instead, an abrupt visual salient point in the interface will direct the user's eyes to the exact physical position of the target cabinet door. The above multi-modal feedback mechanism can reduce the time delay between the system's authorisation and the retrieval of the items.
Finally, the error prompt interface needs to be improved to cope with the problems of the system in non-normal operation. When a user makes an error or cannot pick up their items, a general error message or vague error code will irritate the user. Replace the abstract notifications on the revised interface with specific, context-sensitive operation instructions. List the reasons for the errors, such as code expiry or structural mismatch, and also give a sequence of corrective actions that should be taken after the error. Through this strategic transformation, not only can errors be corrected immediately, but procedural learning will also be improved to maintain the overall usability of the campus food takeaway locker system under various operational circumstances.
5.2 Optimization Schemes for Locker Internal Partition and Spatial Layout
To improve the efficiency of operation and enhance the user's experience, it is necessary to improve the design of the partitions and layout of the campus food pickup lockers. The purpose of industrial engineering's optimization of the space layout is to meet the demand for modularisation, and as such, the actual distribution of the payload size and heat dissipation requirements must be considered in the construction of the physical compartments. At present, campus food pickup still uses a general-purpose type of compartment, and therefore there is a waste of space and no function. To address this, the homogeneous structure will be broken up and replaced with a heterogeneous partition scheme to improve space utilisation and adapt to the various shapes of food distribution.
Based on the reality of campus food pickup, the optimisation plan will change the internal structure of the cabinet to different sizes: small, medium, and large compartments. Smaller compartments will be used to hold beverages and snacks, which are part of campus deliveries but take up very little space. Standard compartment specifications are for a typical type of load—regular takeaway lunch boxes—and have a particular range of heights and widths to avoid vertical stacking damage. Large compartments are used to hold group-sharing meals, and they can meet the logistical needs of collective ordering in student dormitories. Through the tiered structure of the compartments, the waste of empty space in the locker arrangement will be reduced, thereby improving the total capacity for goods at the pickup station per unit area.
At the same time, add a certain amount of constant temperature insulation layers to the partition. The purpose of this measure is to address the problem of insufficient time for food logistics services, which should be considered one of the important factors affecting the quality of food. Due to random delays in the arrival of deliveries and students picking them up, regular bins cannot maintain the serving temperature of heat-sensitive foods. Use constant temperature modules made of localised insulation material and low-power heating elements; the system will actively maintain the sensory properties and food safety standards of the stored items and greatly increase the reliability and satisfaction of the system.
In addition, the operation of compartment allocation should be based on proximity rather than randomness. The optimization of compartment opening distribution will result in the control algorithm of the system assigning the closest idle compartment to the user's scanning position. The above reason is to reduce the amount of physical work that needs to be done by users and relieve the pressure on their minds; therefore, they will be able to go to different parts of the locker system to pick up their belongings. Based on the physical positions of idle compartments and the interactive terminal, the distribution algorithm selects one of the nearest idle compartments in sequence to carry out retrieval operations in the next time period.
Finally, to avoid users having to carry luggage during retrieval, a special temporary platform has been constructed at a suitable height, and it is part of the structure of the locker island. A short-term exhibition platform with good ergonomics will be set up at the same level as the locker island. Users can put their belongings away temporarily on the operation path; thus, their hands will be free to safely and efficiently take food parcels out of the cabinet. This new infrastructure has solved a small problem in the use of the system and will help build a more convenient and easy-to-use food pickup system for the campus.
5.3 Supplementary Design of Campus Scenario-Oriented Auxiliary Service Functions
Add some supplementary designs to campus-scenario-oriented Auxiliary Service Functions to better support the core improvement plan of campus food pickup lockers, that is to say, improve the availability of campus food pickup lockers. The basic concept of auxiliary service functions in Industrial Engineering is that it is the secondary part of the first human-machine interaction system, and it is designed to overcome the difficulties of the context and improve the efficiency of the entire service. The basic concept of the above design phase is that it should adapt to the situation; that is to say, when it comes to campus food delivery, the operating parameters of the locker system should be adjusted according to the different characteristics of time and space and the needs of many students. By putting the above-mentioned assistance measures in place, the system has changed from passive storage to active participation in campus logistics, and thus the asymmetry of information will be lessened and the efficiency of the flow of people will be improved.
To address the problem of severe congestion during concentrated peak pickup hours, a peak pickup queuing reminder function has been added. This function is based on a continuous data synchronization pathway between the embedded sensors of the locker and the campus takeaway mini program. To carry out this operation, the occupancy rate of the nearby area and the number of people waiting to pick up in the near vicinity will be collected in real-time and sent to the cloud server. Then the mini program will display the above data to the people who will use it in the future, so they can decide on a suitable time and place for themselves in advance. From the perspective of industrial engineering, the above mechanism is very important; it acts as a decentralised demand management tool that can smooth out the peak demand curve, avoid local queue congestion, and thus ensure that the system can continue to operate at full capacity during peak loads.
Based on the flexible working hours of the campus users and the scattered time schedules, an automatic food overdue pickup notification has been added. The basic idea here is to actively manage the life cycle of the service. The operation process will be divided into two levels of escalation. First of all, if the system detects a time deviation that exceeds the set time limit for collection, it will send a primary reminder through the in-app message of the platform to gently nudge, as it has low marginal cost and high penetration. If it still hasn't been retrieved in two hours, the system will use a short message to inform the device. The purpose of this redundant communication path is to guarantee that the message can be delivered under different network environments and application active states. A two-tiered architecture is important for actual use; it can shorten the time that food is stored in the inventory, reduce the possibility of spoilage, and recover the space available for lockers, thus improving the utilisation rate of the physical facilities directly.
Finally, to deal with the problem of the disconnection between the staff of the food delivery company and the users due to physical separation, add a one-click contact button on the operation interface of the locker's touch screen. The implementation path is to build a safe, system-driven VoIP bridge that will be activated when the QR code of collection is scanned or the pickup code is entered. The design complies with the principles of privacy by design and hides the personal mobile number for fast voice coordination. Given the complicated distribution environment on campus, there are often unclear drop-off instructions and insufficient storage space in compartments for delivery staff; at the same time, users may get lost or be unable to find the lockers. The one-click interface will help us reduce the time that delivery vehicles stay in place and shorten the time required for users to pick up their packages. Finally, the above additional functions will work together to solve the problems peculiar to the campus environment and restructure the locker system as a more convenient and user-friendly logistics terminal.
5.4 Field Validation of Improved Prototypes and Comparative Test of Usability Performance
5.4.1 Construction of Improved Functional Prototype Based on Test Results
In other words, the new functional prototype has been tested on the ground. Prototype construction in engineering design is not only about producing a physical model; it also embodies the principle of user-centered design, which is to convert abstract ergonomic parameters and interaction logic into a concrete operating system. For the campus food collection lockers, based on the above reasons, the campus food collection lockers will be equipped with new functions to meet the needs of people who live alone, strengthen the interaction with campus life, and satisfy the demands of daily life.
The construction of the improved prototype began by choosing a baseline unit. A common and widely used campus food pickup locker was selected as the first Hardware platform to conduct the subsequent comparative tests. To change the layout of the partition. Based on the above research, we can reorganise the structure of the space for the new campus food collection lockers. The small and medium-sized parcels with high frequency have been placed in the newly adjusted shallow compartments to reduce the physical reach required for users; at the same time, the low-frequency, bulky items have been put in deep Volume compartments located at ergonomically accessible Vertical strata. The reorganisation of the space can solve the problem of too many compartments that cause stress to people's bodies and slow down their work.
Then the human-computer interaction interface has been redesigned for all functions. A large, high-contrast touchscreen with an easy-to-operate Graphic User Interface (GUI) has replaced the old monochrome keypad. The new HCI system implements the main principles of cognitive engineering; it has larger touch targets that match the size of fingers and fingertips, reduces the cognitive load by means of hierarchical task sequences, and has clear visual feedback loops that let users know when a code has been correctly recognised and the door has been opened. In addition, there is a dynamic error recovery path; when the authentication code has expired or been entered incorrectly, it will direct the user to take corrective action and shorten the waiting time at the terminal, reducing the user's frustration.
Other minor service functions were also added to solve some other small usability problems. A new integrated ambient notification system has been installed to help find the right compartment in a high-density retrieval environment; it uses directional LED arrays and localised sounds to guide users from the scanning input terminal to the corresponding brightened compartment. A contactless sanitation station and a temporary parcel-staging shelf were also added to the perimeter of the locker to carry out the related service function strategies that improve the overall user experience after picking up the goods.
To guarantee the reliability and internal validity of the new field validation, a strict pre-operation test was conducted on the constructed prototype. The verification protocol was divided into several iterative stages. At the beginning of the first round, a static functional inspection was carried out to confirm that the structure of the modified partitions was intact and that the hardware for the interactive screen and peripherals had been installed properly. The second round of testing was to assess the dynamic operation of the system, and simulate pickup scenarios in various compartment types to verify that the HCI logic can operate normally, the door can be opened normally and ambient notifications will be sent correctly. The third round is continuous operation stress testing to prevent system failures caused by software lag or sensor out-of-sync. Only after passing all the above strict pre-operation checks and confirming that all the optimised functions have worked normally and met the expected design requirements, the improved functional prototype will be eligible for official empirical field testing and comparative usability evaluation. Based on the above careful construction and verification work, it can be ensured that any differences in performance observed in subsequent field tests are due to the actualised usability improvements and not other mechanical or software problems.
5.4.2 Comparative Test of Usability Indicators Before and After Optimization
The first is to conduct the comparative test of usability indicators before and after optimisation, which is the empirical part of this study and is used to verify the effect of the newly proposed design. In short, in comparison to traditional methods that only carry out descriptions, a quasi-experimental method is now employed in the study of this paper. The basic idea is that all other things are equal; that is, after isolating the changes in the design alone, any change in the independent variable of usability can be considered a result of the improvement plan rather than other interfering factors. At the same time, in the application, such a detailed comparison is necessary; otherwise, it cannot be an objective measure of user-friendly characteristics, and thus, it cannot provide strong support for the continuous improvement of products and industry standards.
The operation plan has been carefully drafted to improve internal credibility. The same group of sample participants was used to conduct the interaction with the unmodified original campus food pickup lockers and the improved functional prototype. A within-subjects design will be used to eliminate variations in individual cognitive processing speed, technological skills and spatial sense. To reduce the impact of a sequence effect and learning effect, in which having used the original system beforehand might artificially increase the performance of the improved prototype, the order of the test sessions was changed for each participant. All field tests were carried out in the same environment to control for the ambient light, acoustic noise, and layout of the space; thus, the ecological validity of the campus was maintained, and other conditions were not altered.
The system has been logging the main usability indicator data of every pickup transaction for a certain period of time. The three dependent variables that will be compared are the average single pickup operation time, misoperation rate, and user operation success rate for the first time. The average single pickup operation time is an indicator of operational efficiency; it is the time taken from a user's first login to the successful opening of the compartment door. The misoperation rate refers to the number of procedural errors, such as wrong screen entries or attempts to open a wrong compartment, etc., and thus measures how well the system's operation logic matches the user's mental model. The first-time user operation success rate is the proportion of users who successfully complete the pickup process for the first time and do not need to be corrected.
In order to go beyond simple descriptive statistics and explore the actual effect of the renovation, some inferential statistics were used. For the operation time data in the continuous ratio form and the nominal form of success and misoperation rates, paired-sample t-tests and McNemar's tests were used respectively to determine whether there were differences in the data before and after optimization. The level of significance has been set to an alpha of 0.05. The analysis is aimed at determining whether there is a significant decrease in average operation time and misoperation rate and a substantial increase in first-time success rate. Based on the exact p-values and effect sizes of the main indicators, the degree and reliability of the usability improvements were determined, and the theoretical basis for the targeted improvement measures was empirically verified. Therefore, a systematic analysis is carried out to ensure that the optimized locker prototype is statistically sound in terms of industrial engineering usability and does not appear to be due to random fluctuations.
5.4.3 Evaluation of Comprehensive Improvement Effect and User Feedback Collection
To evaluate the effect of the general improvement, we need to build a multi-dimensional assessment system to measure how much the campus food pickup locker has been enhanced. The main parts of this evaluation are operational operation efficiency, user subjective satisfaction and scenario adaptability. The main objective is operational operation efficiency; that is to reduce the time and cognitive burden of retrieving. Based on the mean of the task completion time and error rate before and after the intervention, the gains in operation will be shown. User subjective satisfaction is about how users feel, which is measured by a standardised usability scale; for example, ease of use and usefulness. Scenario adaptability measures whether the system can meet the needs of users in different environments and operations, such as peak-hour congestion, low light, and various parcel sizes. Combining the three dimensions together can give a comprehensive numerical result of the system's improvement in usability; thus, there will be no loss of efficiency, quality, or environmental protection.
At the same time, strict collection of user feedback is carried out through on-site questionnaires and semi-structured interviews. A Likert-type scale is used in the on-site questionnaire to collect the immediate, quantitative data about the improved prototype's interface intuitiveness and physical interaction quality. Semi-structured interviews are used to collect more qualitative and emotional information that would not show rigid numbers in a questionnaire. With the help of a semi-structured protocol, it is possible to collect some new themes; in other words, people are willing to share the psychological friction they have encountered in the pickup process. This method can better reflect the overall behaviour of a large number of people and be more representative of the small numbers of people's experiences.
Based on the systematic comparison and combination of multidimensional performance data and qualitative user comments, this study has verified that the general effectiveness of the targeted improvement plans is still in line with the original objective, and it has also discovered some small, unaddressed problems with usability that remain in the current version. Observational data and interview records might show that although the new visual indicators can speed up the main scanning task, there are still some unclear details in the secondary confirmation step, or there are some specific physical reach restrictions for users with non-standard body types. These remaining friction points are small-scale problems in the system that have been somewhat reduced by the macro-level redesign but have not been completely removed.
Therefore, to further optimise this part of the code, we need to identify the problems caused by residual defects. The next iterative improvement should be a small-scale parameter adjustment and localised interface calibration rather than a large-scale restructuring. In the future, we can change the degree of visual feedback saliency according to the different situations and make the physical interaction areas of the locker room more flexible to meet the needs of a wide range of people with different body types. At the same time, the optimization path needs to include algorithmic improvements to automatically allocate locker resources according to real-time scenario density, so that the scenario adaptation ability can be improved without adding more operation burden to users. By systematically addressing these micro-level shortcomings, the campus food pickup lockers will be gradually improved in their all-round usability; thus, the next stage of engineering revisions will be based on practical application tests and feedback from users.
5.5 Promotion Paths of Improvement Strategies Adapted to Different Campus Environments
The campus food pickup lockers need to be modified according to the actual conditions at the time, as different parts of the campus have different people, functions, etc. If the above measures are applied uniformly and excessively, it will waste resources and reduce the effect. Therefore, various promotion methods need to be formulated to promote different kinds of usability for different types of campus locations. Selective modification will allow the proposed plans to be applied flexibly in different campus environments at low costs and improve the overall efficiency and user experience.
The small-scale pickup lockers near the dormitories have quite high and predictable traffic; generally, it is crowded at meal times. In this case, there is no reason for the overload of human senses; it is due to the large number of people. Therefore, for these places, we will mainly focus on improving the operation process and adding peak time-diversion functions. Operation flow optimization is to shorten the physical contact time, such as reducing the time people wait in front of the locker for a long time by quickly scanning the QR code and immediately opening the door. At the same time, some peak diversion auxiliary functions can be used to ease the crowding. Provide a dynamic pick-up schedule and give small rewards for off-peak pick-up; show the current traffic status of the site on the mobile app in real time for users before they arrive. Focus on the flow efficiency and shift the demand; then, the problems in the small-scale dormitory area will be fixed without a large investment in building a new locker system.
On the other hand, the pickup lockers near teaching areas have a large number of new, temporary users and are not frequently used. For them, it is relatively difficult to learn and use this system for the first time. Therefore, the promotion path here should focus on reducing the number of steps required to operate the interface and enhance the voice guidance function. To reduce the interface, remove unnecessary graphics and use a simple task-based UI to guide users through the authentication and locker retrieval process in as few steps as possible. Auto-detection of authentication tokens and removal of unnecessary confirmation screens reduce interaction friction. At the same time, the installation of a good voice guide function will provide cognitive support for new users. Step-by-step auditory instructions can help new users learn how to operate the system intuitively without prior knowledge. Therefore, in this way, one does not need to know much about computers to use them.
Finally, the locker sites near faculty residential areas have different life styles and consumption habits; thus, people there tend to order large family-sized meals, and due to work or family, the time for food to be picked up is often longer. The promotion path of this kind of place should focus on the configuration of heat preservation chambers and large-size family meal chambers. To address the problem of temperature loss in the freezer that affects food quality for a long time, add a module with active or passive thermal insulation. At the same time, in order to meet the volume requirements of large family meals, a larger compartment has been added; thus, it is no longer necessary to book many separate compartments or repackage the food on-site. The above-mentioned hardware configuration also fits the ergonomics and quality needs of the staff.
To sum up, systematically mapping the improvement strategies to the particular environmental characteristics of a campus is an important way to spread and apply the usability-enhancing work. By dividing various modes of promotion, the cost of change is relatively low, and changes can be concentrated on the main problems of poor usability in each micro-environment. Therefore, the above adjustments are not only feasible in theory but also suitable for the various situations of the university campus.
Chapter 6 Conclusion
In this paper, the user test of campus food pickup lockers has been carried out empirically, and according to the results, the key conclusions of campus food pickup lockers' service provision for students are obtained. The primary problems of the current locker system, based on field observations and user interaction analysis, are the weak points in usability. Most prominent among the problems is that there is not enough space during the busy period of pickup, and as a result, many people will be waiting in line around the lockers and causing chaos. Due to the interface interaction obstacles, such as the vague notification mode, complicated verification of the retrieval code and lack of guidance for operation, users have more cognitive effort and time delays. The problem is that the compartment depth and height are not suitable for the size of food packaging, so there may be problems of space mismatch, items getting stuck and other issues affecting the operation.
To improve the above problems, some targeted optimization strategies were proposed and verified in practice. According to the results, it can be seen that the usability of campus food pickup lockers has been significantly improved. By redesigning the distribution of space based on the changing flow of people and reducing the number of steps in human-machine interaction, the operation efficiency of retrieval is significantly improved. With clear visual levels and simple operation and feedback modes, users are less likely to make mistakes or hesitate. Therefore, the overall usability index, including effectiveness, efficiency and user satisfaction, shows an upward trend and thus proves that the human-centered design interventions are practical and theoretically sound [12].
According to our research, there are differences in the demand for food pickup services among various campus user groups. The statistical analysis of the field test data shows that students prefer to retrieve the food quickly and access it frequently, and hope to spend the least time possible and have a simple way to use it. Faculty and administrative staff are more tolerant of minor delays in time, but still demand that the compartments are clean, sturdy and the food temperature remains intact. The operational supply-side users are mainly the delivery personnel of the compartments and need an efficient batch loading interface to reduce waiting time when replenishing large quantities of goods. There are different needs for the design, so modular and flexible systems should be used; otherwise, the system would be too rigid.
The present study has some shortcomings, even though the methods used are relatively strict. The number of field test samples is small; although it is enough to draw some initial statistics, the absolute amount is relatively small, and thus the results may not apply to all universities with different infrastructure scales. The time period for collecting empirical data is too short, and there is no basis from long-term operations data. Because of the time limit, it is not possible to carry out a complete longitudinal study on the reliability of the system, degradation of components, and the long-term effects of usability improvements under different seasons and academic cycles.
In the future, we should not only improve the hardware but also develop smart campus management systems to work with it. Advance to an intelligent, self-regulating model for the campus food pickup locker system. Real-time data analysis, prediction algorithms and the Internet of Things can all be used to make the system automatically adjust the allocation of compartments, according to the continuous modification of the user's behaviour modelling by dynamically adjusting interface prompts and proactively arrange maintenance schedules. It will not only overcome the current difficulties in use, but also make the locker system more flexible and improve the service quality of the smart campus [6].
References
[1]王紫玥. 面向校园环境的智能快餐柜设计研究[D]. 武汉理工大学, 2024.
[2]陈晨, 岳凯兴, 张懿. 高校无接触式外卖保温柜的设计[J]. 物联网技术, 2025, 15(5): 118-122.
[3]张培僧, 惠琛云, 李春华. 基于工业一体机控制的智能网联外卖柜[J]. 物联网技术, 2025, 15(4): 68-71+79.
[4]胡雨晴. 无接触智能服务采纳影响因素研究[D]. 浙江财经大学, 2024.
[5]李根, 张亿萍. 物联网驱动的高校智慧食堂创新设计[J]. 中国储运, 2025(8): 174.
[6]王思琪, 温鑫, 汤宗旭. 基于STM32单片机的智能外卖柜存取系统设计[J]. 电子制作, 2024, 32(15): 84-87.
[7]曹京龙, 周艳霞, 李瑞阳. 基于STM32单片机的智能外卖柜设计[J]. 互联网周刊, 2024(13): 28-30.
[8]郭宇. 多渠道履约机制下外卖平台的运营优化策略研究[D]. 中国科学技术大学, 2026.
[9]Wan, Zhong, Wang, Weitong. An optimal takeout supply system of intelligent self-pickup cabinets: Evidence from campus safety during COVID-19 in China[J]. 2023.
[10]Thongkam, Orawee. Factors influencing adoption and usage probability of automated parcel lockers[J]. null.
[11]Cieśla, Maria. Perceived Importance and Quality Attributes of Automated Parcel Lockers Services in Urban Areas[J]. 2023.
[12]Ali, Mir Ihrar. Design and Evaluation of ePub: Enhancing Online Food Ordering for Campus-Based Food Services.[J]. 2025.