T his project concerns the integration of machine learning and sensor-based control in intelligent robotic systems. The research combines techniques of explanation-based control with robust and adaptive nonlinear control, computer vision, and robot motion planning. We wish to go beyond the strict hierarchical control architectures typically used in robotic systems by integrating modeling, dynamics, and control at all levels of intelligence. Our ultimate goal is to combine analytical techniques of nonlinear dynamics and control with artificial intelligence into a single new paradigm, in which symbolic reasoning holds an equal place with differential equation-based modeling and control.

This project concerns the robust and adaptive control of underactuated mechanical systems. This class of systems is quite broad and encompasses flexible structures of all kinds including flexible link robots, flexible joint robots, as well as robot models that include actuator dynamics, and many of the classical control problems like the ball-and-beam and cart-pole systems. The unifying theme of the research is to use the ``divide and conquer'' design philosophy of the recently developed method of integrator backstepping to extend to underactuated systems the powerful design techniques now available for fully actuated systems, such as feedback linearization and passivity based methods. The ultimate goal is to greatly extend the applicability of such nonlinear design techniques.