Session 11: DHM Modeling Methods
Authors: Karim Abdel-Malek (The University of Iowa) , Rajan Bhatt (The University of Iowa) , Chris Murphy (The University of Iowa) , Marco Tena Salais (The University of Iowa)
Human modeling and simulation environments have become sophisticated, providing capabilities for posturing the digital avatar, conducting dynamic tasks, assessing human performance, and obtaining scientific analysis. Physics-based analysis of human motion has also enabled capabilities for predicting the dynamics of the motion whereby the avatar is able to respond to external effects such as load conditions or the physical environment. However, the behavior of the avatar is typically the same. Indeed, the same input to the system yields the same output. This work is concerned with obtaining more realistic and different behaviors for the human while conducting a task in a physics-based environment.
While the Santos1 digital human modeling (DHM) system has been successfully used in the analysis and prediction of human motion, the resulting behavior of a specific soldier is the same. An (avatar) individual doing a task will do the same task exactly the same way if the simulation is executed again. In reality, however, a soldier may adopt a different strategy to accomplish the same task.
This paper discusses the development of a rigorous scientific method using human performance functions as the driving force to induce different behavior.
The Santos platform uses 215 degrees of freedom to model a human. The research and development of this platform over the past 15 years has made significant strides in the development of predictive dynamics2 as a methodology for predicting behavior and is physics-based. Rather than solving the highly complex and coupled algebraic-differential equations of motion governing the human’s behavior, predictive dynamics uses optimization to solve for the behavior.
In this paper we use a new methodology that employs seed scenarios before executing predictive dynamics, thus allowing the system to provide variations in task execution. A tall person will execute a task differently than a short person, if height is an influencing factor.
The results of predictive dynamics are autonomous prediction of the motion while subject to the laws of motion (we use Lagrange’s equations of motion). These results are not pre-recorded but rather a prediction of what a human can do. Depending on the task at hand, the soldier in the DHM environment will accomplish the task unaided by human analysis but subject to physics, biomechanics, physiology, and the constraints of the environment.
This paper will present the methodology for predictive behavior to affect various strategies. Results of this work will be presented.
Keywords: soldier, human performance, predicting behavior, physics-based
How to Cite: Abdel-Malek, K. , Bhatt, R. , Murphy, C. & Tena Salais, M. (2022) “Santos the virtual soldier predicts human behavior”, Proceedings of the 7th International Digital Human Modeling Symposium. 7(1). doi: https://doi.org/10.17077/dhm.31790