Optical Flow Based Obstacle Avoidance and Path Planning for Quadrotor Flight

摘要

This paper demonstrates the use of optical flow for navigating UAV to the destination without collision with obstacles. Optical flow information is utilized in detecting obstacles and the repulsive potential field is generated by gradient vectors and TTC of detected obstacles. One of the key parameters is discovered to be the gain ratio of attraction force to repulsive force. Small ratio will result in longer, more circuitous routes, while appropriate ratio will lead the UAV to pass through the obstacles safely and quickly. We implement our approach in virtual environments and conducted experiments with multiple obstacles. Our method has been demonstrated to be effective for real time obstacle avoidance and path planning.