Adaptive Torque Control of a Robotic Manipulator
Abstract
In this thesis, I present the theory and application of enabling intuitive control and physical interaction between a human operator and a mobile manipulator. To do so, the model of the manipulator was estimated for its kinematic and inertial properties. Specifically selected in this project was the Kuka youBot for its ease of use and wide academic availability. With thorough modeling of this device that included the arm as well as the base, and advanced control and guidance techniques, an adaptive controller was used to guide the manipulator towards a goal trajectory with great accuracy. The goal trajectory was built in cartesian reference frame was translated into the n-dimension joint parameters of the device through the kinematic techniques here discussed. These were then demonstrated with thorough virtual simulation using Matlab Simulink, and also applied to an actual device using custom ROS code written in C++. These demonstrations highlighted the difficulty of cartesian guidance with a limited manipulator whilst also showing adaptive methods can work with jacobian based error estimation. While the jacobian math suggests orientation as well as position guidance are possible, there are limiting interactions in under articulated manipulators that can be overcome using motions of the mobile base. However, until these base motions can be well observed, they cannot be used with an adaptive torque control of a mobile manipulator.