Abstract: A new obstacle avoidance algorithm based on Dubins path was presented for unmanned aerial vehicle (UAV). Considering the aviation performances and the least swerve radius of UAV, adopting genetic algorithm (GA), a path planning method was established to get the optimal path while all the obstacles' positions were known. The algorithm was improved so as to solve the problems that the obstacles' positions were still unknown, and made the UAV have the ability of real-time obstacle avoidance. The simulation results showed that the principles of these algorithms are correct, and UAV can be guided to the intended target reliably with these methods shown in this article.