Symbolically Synthesized Motion Primitives for Autonomous Navigation
Zhixin Zhao, Antoine Girard, Sorin Olaru · 2024
This paper proposes a novel approach to navigation for autonomous vehicles that leverages symbolic control methods and system translational and rotational invariance properties. By decomposing in-plane motions into translations and rotations, the approach constructs corresponding motion primitives that enable efficient offline controller design and avoid computationally expensive discretization of the whole state space. The resulting controllers achieve complex trajectories through concatenation of motion primitives. At the same time the safe corridor given by this method will provide safety guarantee for the whole mission.