Zoom: https://upenn.zoom.us/j/3458894777?omn=91979093447
Meeting ID: 345 889 4777

Physical compliance enters robotic systems either by design, to yield gracefully to disturbances, or by circumstance, as a property of the environment or task itself. Often this compliance is distributed across many interacting elements, and how those elements are connected determines whether their combined effect on the robot’s configuration can be resolved locally or as a coupled system. In this talk, I will discuss how to model and plan over two such systems: a soft manipulator arm with tunable stiffness elements, and a physically connected team of mobile robots that cooperatively traverse deformable, slip-prone terrain. We use a graph-theoretic model of the connected compliant elements in each system to capture how motion is coupled to the underlying forces. Leveraging its model, the soft arm motion planner adjusts internal stiffness to redistribute forces resulting in a target shape, while for the connected robot team, where terrain compliance is fixed, the planner reasons over modeled interaction forces to select robot actions that guide the team toward a goal. In each case, the same underlying model supports planning across a range of network topologies and terrain conditions.