CARTESIAN MOTION CONTROL OF THE AUTONOMOUS MOBILE ROBOT MACROBE

Autonomous mobile robot research at the Laboratory for Automatic Control Engineering of the Technical University of Munich focuses on multisensory data acquisition and information processing in planning, navigation and vehicle control. For purposes of validation and demonstration the robot vehicle MACROBE (Mobile and Autonomous Computer controlled ROBot Experiment) is developed. This paper outlines the overall architecture of MACROBE's information processing system and discusses in more detail the newly developed Cartesian motion control subsystem. This system is applicable to any vehicle with tricycle kinematics. It has a comfortable and open user interface and provides high path accuracy. The robot's 2D motion path is specified by a sequence of Cartesian subgoals. In a first step, the local path planning interpolates the subgoals by a combination of line and arc segments. In a subsequent step, a path with continuous curvature is obtained through feedforward and feedback control and approximation of the arc segments by polar splines. Polar splines allow the calculation of the required steering angle based on the actual vehicle posture. This paper discusses algorithms, implementation and experimental results and compares the accuracy of the motion control and the dead reckoning system.