The interaction of force and geometry in robotic manipulation

Robots are used exhaustively for contact tasks such as grinding, scribing, and deburring in the manufacturing industry. In these tasks, the robot has position constraints in some directions and force constraints in other directions. It is most often the case that the robot is commanded to maintain a contact force in the normal direction of the workpiece surface while following a trajectory in the tangent plane. For the robot to successfully maintain contact while performing the task, it is essential to know the precise position and geometry of the workpiece, but in practice, there are uncertainties in both. Inaccuracies in workpiece mounting combined with manufacturing tolerances and workpiece wear significantly deviate the positioning and part information from the ideal image of where the workpiece should be and how it should look. The following dissertation develops a novel process that allows the robot to complete successful contact tasks during the presence of workpiece uncertainties and workpiece movements. Furthermore, a strategy is devised that uses existing sensors on industrial robots to gather information to update workpiece geometry and placement in real time, thereby reducing the workpiece uncertainties and improving the overall performance of the robot. A number of simulation experiments demonstrate that these methodologies can be successfully applied to a typical industrial robot with 6 degrees-of-freedom.