Development and testing of an impedance controller on an anthropomorphic robot for extreme environment operations

Abstract

This thesis explores the use of impedance control on an anthropomorphic robot for operations in extreme, poorly mapped environments. First, a dynamic model was developed for a Baxter Research Robot. This model improved on standard dynamic models for similar robots by including the dynamics of the actuators in the system. Specifically, it was demonstrated that when the effective inertia of the actuators is neglected, the system will transmit 1.6 times more force to the environment than the model predicts. A force based Cartesian impedance controller was then implemented on Baxter, and numerous ways to modulate the endpoint impedance, including feedback and geometric configuration, were discussed and compared. Finally, a series of scaled down tasks similar to ones which are required in the decommissioning of offshore oil fields were then completed on Baxter using the Cartesian impedance controller. Overall, it was demonstrated that by using this more advanced control scheme, Baxter was (1) capable of satisfactorily completing the scaled down tasks, (2) more robust against errors in the map of the environment than with traditional controllers, and (3) capable of improving the map of its environment while completing the task.