Human-Machine Interface Design of Construction Machinery Based on Scenario Generation and Accident Pattern Analysis

Abstract

Background The interaction between construction machinery and operators directly impacts both work efficiency and safety, making user-centered design increasingly important in the development of Human-Machine Interfaces (HMI) for construction machinery. In recent years, the construction machinery industry has been actively advancing the development of cutting-edge equipment such as unmanned construction machines and remote-controlled machinery. However, as long as human involvement persists on-site, it is difficult to completely eliminate the limitations and potential for human error. Therefore, it is crucial to be vigilant regarding errors that may arise from human involvement and to identify patterns in construction machinery accidents. The aim of this study is to minimize human error in the operation of construction machinery by analyzing accident patterns, designing scenarios, constructing virtual simulations, and verifying the proposed accident scenarios through expert evaluation. Methods Accident cases involving construction machinery were analyzed using a human error model to identify recurring human error factors. These factors were then visualized in graphical form to analyze the patterns and data among accident causes. Subsequently, by training a generative AI model with accident cases and contributing factors, accident scenarios specific to each type of construction machinery were generated, emphasizing the importance of deriving realistic and detailed scenarios. The generated scenarios were then filtered based on the frequency of contributing factors to select representative accident scenarios for each type of construction machinery. These scenarios were further developed into simulations that replicate situations operators may face, and were validated through user testing. Results The results of the simulation experiments, based on representative scenarios for different types of construction machinery, revealed varying responses among participants when asked about their experiences with and utilization of icons and sounds on the windshield display. In in-depth interview questions regarding the similarity of the simulation to real-world situations, most participants reported that the simulations closely resembled actual scenarios, representing frequent situations where accidents are likely to occur. Based on the generated scenarios, a windshield display design concept was developed, and experiments were conducted to gather various opinions and reactions from participants. Throughout the experiment, participants generally provided positive feedback on the design concept, expressed significant interest, and demonstrated a high intention to use the proposed concept. Conclusions This study confirmed that the design concept based on the generated scenarios could be practically applied to HMI design for construction machinery. Moreover, the identification of predictable accident scenarios for each type of construction machinery in the research and industrial fields suggests the potential for proposing a significant direction for future HMI design development in this domain. This study demonstrated how user-centered design methods can be applied to HMI design in the construction machinery field, a domain that has traditionally received less attention in design. Scenario-based approaches grounded in data not only contribute to the scientific rigor of the design process but also show potential for application in any field where realistic simulations are utilized.