Motion control for omni-drive servicerobots under Kinematic, Dynamic And Shape Constraints

Abstract

In this paper, a fast reactive obstacle avoidance approach for omnidirectional driving is presented. The method is based on the dynamic window approach, but uses a cuboid instead of a window to limit the 3-dimensional search space accordingly to the dynamic constraints of the robot. Besides the kinematics and dynamics, the presented approach also considers the shape of a robot. To cope with the effort of the time consuming distance calculations, the remaining distance values are precalculated in an offline part and provided by a lookup table. This procedure is based on the Curvature Distance Lookup (CDL) approach which is extensively used in several real world robotic applications but which was so far only implemented for 2-DOFs. During the online phase, the extended approach enables the selection of a motion command from a wide range of curvatures (3-DOFs) within the current dynamic cuboid. The distance values are queried from the corresponding lookup table entries depending on the occupancy grid determined by latest sensor information. The reduced computational effort of the control loop allows to consider obstacle information from corresponding sources to the full extent and without preprocessing. Furthermore, complex heuristics can be implemented to evaluate a high number of omni-drive velocity triples in such a way that the driving behavior of the robot is influenced accordingly.