Optimization-Based Prediction of a Soldier’s Motion: Stand-Prone-Aim Task

The objective of this study is to simulate the Stand-to-Prone-to-Aim task of a soldier using a full-body, three dimensional digital human model. The digital human is modeled as a 55 degree of freedom branched robot. Six degrees of freedom represent the orientation and position of the pelvis coordinate frame of the digital human and 49 degrees of freedom represent the revolute joints which model the human joints. Motion is generated by a multi-objective optimization approach minimizing the mechanical energy and joint discomfort simultaneously. The optimization problem is subject to constraints which represent the limitations of the environment, the digital human model and the motion task. Design variables are the joint angle profiles. Using the given method, we can predict a realistic motion for the “Stand-to-Prone-to-Aim” task. We are also able to very well predict the “Natural Point of Aim” for this task which is validated (along with other determinants of motion) and matches very closely with the experimental data which is motion captured in the VSR Lab at the University of Iowa. Inevitable transfers of weapon as an external object between the two hands have to occur in performing this task. Collision avoidance of the rifle with the hands and body during these rifle transfers is a very challenging constraint that has been implemented. These collision avoidance modules use compound primitives such as finite cylinders and finite planes whose edges are smoothed out in order to have continuous gradients for the collision avoidance constraint.