A new trajectory-based path planning approach for differential drive vehicles

Abstract

A new robot local path planning algorithm is proposed that attempts to improve drive path contouring when in autonomous mode, by integrating a vehicle dynamics model into decision making. The technique is formulated in a space made up of path trajectories, where each trajectory corresponds to selected values for four parameters - initial and final forward and turn velocities for the robot. A criterion function that balances speed vs. safety is formulated to pick the best target forward and turn velocities for any given robot state. An advantage of the method is that driving decision commands are directly compatible with typical differential-drive motor controller inputs. The method is implemented on a differential drive robot with torque-limited motors, that is driven in a roadway type environment with occasional partial obstruction. Experimental results are presented.