In this paper, we implement a path planning algorithm and combine with an obstacle avoidance system for a small commercial quadrotor vehicle. To keep the flight safe, the quadrotor navigates with the vision information for obstacle detection. A sequence of images are captured from the onboard camera and used for visual guidance. The optical flow is computed from the images and the derived depth cues are used to detect incoming obstacles. We implement the algorithm on an embedded system and use a single-board computer as the control platform. The experiments carried out in the outdoor environment show the feasibility of our technique for path planning and autonomous navigation.
展开▼
