Path planning for quadrotors in 3D dynamic environments

Abstract

A topic of interest on unmanned aerial vehicles is determining appropriate paths that allow them to move from an initial position to a target position, ensuring that the path is safe, in other words, that there is no risk of collision. In this article, two techniques of path planning and an obstacle avoidance strategy for quadricopters are presented. Both techniques are functional on three-dimensional environments with static or dynamic obstacles restricted to constant speeds. The techniques work on an environment modeling with planes which generate an artificial potential field. The first technique is based on moving points that connect the initial position to the goal, and then every point moves towards free zones of influence of obstacles along the potential field, which makes unobstructed paths. The second technique uses the concept of safe areas, which is used as a criterion for updating the position of the points. Additionally, a methodology of dynamic obstacles avoidance is proposed, which is to transform the problem of trajectory planning with dynamic obstacles to the case of path planning with static obstacles through a process of analyzing the possibility of a collision. The results show that these techniques overcome the drawbacks of the gradient descent-based algorithms as local minima and unstable oscillations problems.