A redundancy-based approach for visual navigation with collision avoidance

We propose an autonomous vehicle guidance framework which combines visual navigation with simultaneous obstacle avoidance. The method was originally designed in, but real outdoor experiments and major improvements have been added in this paper. Kinematic redundancy guarantees that obstacle avoidance and navigation are achieved concurrently. The two tasks are realized both in an obstacle-free and in a dangerous context, and the control law is smoothened in between. The experiments show that with our method, the vehicle can replay a taught visual path while avoiding collisions.