Dynamic Obstacle Avoidance using Online Trajectory Time-Scaling and Local Replanning

Ran Zhao (1) , Daniel Sidobre (1)

1 LAAS-RIS - Équipe Robotique et InteractionS

In various circumstances, planning at trajectory level is very useful to generate flexible collision-free motions for autonomous robots, especially when the system interacts with humans or human environment. This paper presents a simple and fast obstacle avoidance algorithm that operates at the trajectory level in real-time. The algorithm uses the Velocity Obstacle to obtain the boundary conditions required to avoid a dynamic obstacle, and then adjust the time evolution using the non-linear trajectory time-scaling scheme. A trajectory local replanning method is applied to make a detour when the static obstacles block the advance path of the robot, which leads to failure of implementing time-scaling approach. Cubic polynomial functions are used to describe trajectories, which brings sufficient flexibility in terms of providing higher order smoothness. We applied this algorithm on reaching tasks for a mobile robot. Simulation results demonstrate that the technique can generate collision-free motion in real time.

Mots clés: obstacle avoidance; velocity obstacles; trajectory time-scaling; local replanning

Domaines: Robotique [cs.RO]

Ran Zhao, Daniel Sidobre. Dynamic Obstacle Avoidance using Online Trajectory Time-Scaling and Local Replanning. Informatics in Control, Automation and Robotics (ICINCO), Jul 2015, Colmar, France. pp.414-421. ⟨hal-01764018⟩