The objective of this project is to study the control of articulated objects performing complex actions. Motion analysis will be approached from a computational perspective viewing actions as directed by coordinated processes that specify torques and forces on objects to accomplish purposive behavior. The project will study the problems that arise when objects have many degrees of freedom, when tasks lack precise specification, when unpredictable events and interactions with other objects are commonplace, and when the consequences of actions cannot be precisely determined before the action is performed. The project will emphasize fundamental principles for programming complex physical object. Methods for representing objects, specifying tasks, and reasoning about actions will be addressed. Techniques for adaptive, event driven control of concurrent motor functions will be investigated. Particular emphasis will be placed on coordination and methods for maintaining stability and balance. The work will contribute to the ongoing design of a system in which theories of action and control can be easily posed as computational processes and empirically tested in simulation. The physical object simulator Newton will be used to accomplish the work.
