Dynamics and Control of Redundant Robots

Abstract

There is a growing number of robotic applications in which devices with many degrees of freedom dynamically interact with a complex environment. Designing suitable real-time control systems often requires problem specific methods of motion generation, which in turn rely on efficient methods for solving kinematic and kinetic problems. Analyzing such real-time planning and control systems itself requires efficient methods for simulating a multibody system in its interaction with the environment. This habilitation thesis describes different approaches towards controlling and simulating such robots, based on a unified presentation of basic kinematics and dynamics. The focus is on fast simulation of the dynamics of multibody systems, efficient geometry representation for real-time collision avoidance and methods for real-time optimal redundancy resolution. The thesis shows application examples to a redundant manipulator for selective harvesting and to a bipedal walking robot.