Nonholonomic stratified motion planning along decomposed reference trajectories

This work studies a new stratified motion planning algorithm. The proposed technique is able to lead a stratified system along a restricted class of reference trajectories while the earlier approach aimed only to reach a final state without influencing the trajectory. The suitable class of reference trajectories is still restricted because they should go through a sequence of connected integral submanifolds. A highly oscillatory sequence (HOS) of inputs is built in the method to produce a sequence of trajectories converging to the reference trajectory. The major benefit of the new concept in comparison with the conventional approach emerges when obstacle avoidance is required in the configuration space. The method is demonstrated on a simple model of a hexapod robot.