Differential evolution based path planning for UAVs surveillance in wind

The Dubins model is used to describe the dynamics of UAVs in the paper. We focus on the problem of generating optimal paths for UAVs performing surveillance of multiple ground targets, which is modeled as a Dubins Traveling Salesman Problem (DTSP). The computation of Dubins distance between two configurations is simplified through freeing the heading of terminal point, which can greatly reduce the optimization scale and difficulty of the whole problem. Differential evolution algorithm based on random key encoding is used to optimize the visiting sequence of targets, and the suboptimal solution to DTSP can be derived. In view of the fact that environmental factors (e.g., wind) can make great impact on UAVs' path, a virtual moving target is introduced to approximate the influence of wind on UAVs' path. The uniqueness of the minimal time solution in the sense of terminal heading relaxation in wind is demonstrated in theory, and the method to calculate the solution is given at the same time. In the simulation, we verify the effectiveness of the model and make a comparison between the different planning results with wind and without wind.