Jeff M. Wendlandt, Graduate Student
(Professor S. Shankar Sastry)
We are developing control and simulation algorithms for multibody systems to enhance the design of controlled mechanical systems. We are working with the Soldier Systems Army Research Laboratory in Natick, MA to develop predictive, dynamic models of human motion. We are developing controllers to make the human model stand, walk, run, jump, and adapt to loads. In [1], we developed a model-based workspace controller utilizing recursive multibody algorithms to balance the multibody model of a human. The workspace controller is shown balancing the human model in Figure 1.
The system in Figure 1 was simulated with a research multibody simulator with an impulsive contact model. The impulsive contact model is not well suited to simulate the type of contact in the human model. After surveying commercial multibody simulators, we realized that there were no simulators that could efficiently simulate the controlled system. We are currently using the techniques developed in [2] to create an implicit, linear-time multibody simulation platform so that we can continue to design and develop advanced control algorithms. Future work involves designing hybrid controllers to make the model walk, run, jump, change direction, and adapt to loads.