Search papers, labs, and topics across Lattice.
The paper introduces FCFAS-planner, a novel asynchronous swarm planning method designed to address collision risks and planning challenges in unknown environments for ground robots. It combines a multi-constraint path point generation framework with trajectory optimization using a "spring-damping" system and convex decomposition to improve local target point generation. The method introduces a triangular trajectory conflict model and a velocity control method based on the "spring-damping" system to enable robots to avoid collisions by accelerating or decelerating.
Achieve collision-free swarm planning in dense, unknown environments by enabling robots to intelligently accelerate or decelerate based on a novel triangular trajectory conflict model.
In swarm planning for ground robots, the risk of collisions due to trajectory intersections and the challenges of planning in unknown environments are critical issues. To address these challenges, this paper presents a swarm planning method based on asynchronous planning to achieve collision-free swarm planning in dense unknown environments. Path points are generated based on a multi-constraint framework, and trajectory optimization is performed by combining a “spring-damping” system with convex decomposition, thereby improving the method of generating local target points. To avoid trajectory conflicts between robots, we introduce a triangular trajectory conflict model and establish a velocity control method based on the “spring-damping” system. This enables two robots whose planned trajectories conflict to either accelerate simultaneously to escape or decelerate to avoid a collision. Simulation-based and real-world experiments show that the proposed method can generate executable, collision-free trajectories at a relatively high speed in unknown environments while effectively preventing collisions between robots during swarm planning.
