Turn around Behavior Generation and Execution for Unmanned Ground Vehicles Operating in Rough Terrain

We present an approach for turn around behavior generation and execution for an unmanned ground vehicle (UGV) operating in a rugged terrain. The generated behavior allows the vehicle to rapidly execute N-point turn maneuver in a narrow space by exploiting the vehicle’s dynamics as well as the properties of the terrain. The developed approach incorporates model-predictive planning and optimization into the behavior execution. It combines offline synthesis of an atomic maneuver model for a given vehicle and terrain model, fast online computation of motion goals determining the execution sequence of atomic maneuvers and their extent in the state space, and atomic maneuver selection and adaptation to the terrain shape. We have successfully evaluated the turn around behavior using an augmented reality setup by having the vehicle execute 5-point turn maneuver on an inclined platform.