Abstract
In this paper, a trajectory planning method of automobiles for collision avoidance by extending the hydrodynamic potential to dynamic environment using local sensor data is proposed. This potential approach for guiding mobile robots has already been proposed in our previous research. It has no maximum or minimum points and can generate the smooth trajectory in real time that is represented by velocity vectors directly for collision avoidance. This previous method cannot consider vehicle dynamics and local sensor information. To cope with the difficulty, a vortex field is calculated by using a velocity of the host vehicle and distance to obstacles. Simulation results for constructing potential fields using both virtual data including a host vehicle's position with lane information and sensor data measured by a laser range finder on a small vehicle show that the proposed method can generate smooth avoidance trajectories that can be traced by automobiles with vehicle dynamics.
