The Virtual Soldier Research program (VSR) is a research group within the University of Iowa Center for Computer-Aided Design (CCAD). VSR was founded by Professor Karim Abdel-Malek in 2003 through external funding from the US Army Tank Automotive Command (TACOM) to put the Warfighter at the center of US Army product designs. Professor Abdel-Malek's background in robotics and the use of rigorous mathematical formulations was the first introduction of mathematical kinematics to the field of Digital Human Modeling (DHM). Prior to 2003, all DHM models were based on experimental data that use lookup tables to enable the posturing of simple mannequins. Indeed, the first version of Santos, presented at the a DHM conference was met with great success because it was the first fully articulated digital human model that behaved as humans do, whereby joints had constraints (also called ranges of motion) and a user could pull on an arm for example and as a result the entire body would respond accordingly. Cost functions representing human performance measures were used to drive the motion within the optimization formulation. Seated posture prediction for example was accomplished by simply providing the seat geometry. The posture prediction methodology was subsequently validated Later on, the Predictive Dynamics method was created and used the same optimization technique with the addition of 3D laws of motion (equations of motion). The Santos system includes many aspects of physiology modeling, thermal, hand model, grasp prediction, gait analysis including stability, mobility, suitability, survivability, maintainability, training, and many other metrics typically used in the assessment of human performance for the Warfighter.
Using this initial research and funding as a foundation, the VSR program continues to develop new technologies in digital human modeling and simulation, specifically with applications for the military, automotive, manufacturing, athletics, injury prediction, With over $60M in funding, the Santos simulation platform provides two key components that differentiates it among all other DHM systems: (1) Physics based and (2) Predicts behavior.
VSR's digital human model, Santos (R), stands at the center of its digital human modeling and simulation research. The high-fidelity, biomechanically and biofidelic accurate musculoskeletal model incorporates 215 degrees of freedom, including the hand, feet, and eyes. The dimensions of the skeleton are mutable, able to represent any anthropometric cross section. In addition, Santos includes a muscular system with the ability to predict muscle activation and muscle forces in real time, using a novel optimization-based methodology. This method, developed over a period of eight years by the Virtual Soldier Research program is called Predictive Dynamics and published by a book and a large number of papers. Furthermore, the gradient based methodology used to solve for the motion was also replaced with an artificial intelligence neural network method.
The mathematical model for the Santos skeleton was developed based on the Denavit-Hartenberg method for kinematic and dynamic analysis. Optimization is used to determine postures and motions that are governed by various human performance measures (objective functions) and constrained by the restrictions imposed by the skeleton, the laws of physics, and the environment. The Santos simulation platform is being used by the US Military, industry (for example automotive industry), and academia. The Virtual Soldier Research team transitioned a product from the Santos environment called Enhanced Technologies for the Optimization of Warfighter Load (ETOWL) funded by the Office of Naval Research (ONR) to the US Marines. The product was later renamed as GruntSim. This human modeling and simulation environment is now being used to study human factors and ergonomics in many applications. This model includes not only Predictive Dynamics but also stability criteria called Zero Moment Point. The use of the Santos Digital Human Model for example in assessing assembly issues in the design stage have been demonstrated and for gait prediction as well. Running for example, which substantially more difficult to predict for a virtual human, was accomplished.