Simulating Natural Human Performance in Digital Human Model

Abstract

Digital human models (DHMs) represent humans in a virtual environment and are widely used across various fields. In engineering design, DHMs are employed to analyze products or workplace ergonomically during the early design phases by simulating interactions between DHMs and CAD models. Despite advancements in computing power, DHMs still struggle to fully capture the complexity and variability of human anatomy and movement, resulting in less accurate simulations of real-world scenarios. While DHMs are primarily used to assess the risk of work-related musculoskeletal disorders, some literature has explored their use in identifying workplace accident risks. Variations in human performance from the norm can lead to unexpected events that may harm man-machine systems. To identify such risks using DHMs, it is essential to incorporate natural human performance variations. These variations can occur at the decision-making or the action (I.e., task execution) level. This thesis focuses on variations at the action level that can be integrated into DHMs. We developed theories to demonstrate the issues with current methods of providing force and posture inputs and subsequently created frameworks and mathematical models to integrate into the DHM system. Firstly, a coupled object-human behaviour model is presented in a Digital Human Modelling (DHM) environment. This model argues that the responsive co-evolution of object and human configurations is crucial for realistic performance assessment. A method is proposed to model and incorporate the environment’s response in the virtual world. The DHM’s posture is ensured to maintain stability and effort-budget conditions, while objects adhere to the laws of mechanics. This combination reveals realistic behaviour during simulations, highlighting the significance of the proposed environment modelling method. The proposed approach extends the prevailing simulation paradigm by enabling realistic interactions between objects and DHMs. Next, the issue with the current approach to providing force and posture in DHMs is identified. At present, users input force and posture independently, which leads to inaccurate joint stress assessments due to incorrect force direction estimations. We conducted experiments to identify variables influencing the applied force direction and developed a mathematical force model correlating the force direction with the application point and magnitude. Integrating this force model into an existing DHM eliminates the need for manual force direction input. We then focused on variations due to human physiology, specifically muscle fatigue. Muscle fatigue and recovery are physiological processes that impact performance and endurance. Ergonomic task analysis can balance well-being and overall performance of workers. However, most DHM evaluations use static postures, fixed external loads, and unchanged joint strength. By incorporating fatigue and recovery phenomena, we can enhance the quality of the analysis. In reality, posture can vary, and joint strength may decrease due to fatigue, thereby affecting performance. The proposed fatigue and recovery models are integrated into the DHM. Stimulating various scenarios with different loads, work-rest intervals, and load division, it was observed that reducing load and increasing work-rest time improved endurance. However, under time constraints, increased load divisions led to faster work speeds and increased fatigue, highlighting the importance of considering body inertia in load division analysis. The theories and models were integrated into the software using the in-house developed DHM, Maya-Manav, to simulate natural variations. In summary, this thesis develops theories and models to incorporate variations in human performance within Digital Human Models (DHMs). A coupled object-human behaviour model is introduced to enable interactions between DHMs and their environment. A mathematical force model is developed to address inaccuracies in force and posture input, improving joint stress assessments. Fatigue and recovery models are integrated to evaluate endurance and performance under different conditions. These methods are implemented in the in-house DHM software, Maya-Manav, to analyse risks arising from variations in human performance.