Improved Heuristic Algorithms for UAVs Path Planning in Hazardous Environment

Abstract

As the route planning of UAV searching in a risky environment is a complicated combinatorial optimization problem, which is characterized by a variety of unpredictable factors. Heuristic methods can be used to speed up the process of finding a satisfactory solution. In this paper, Greedy algorithm and Q-learning algorithm are designed to efficiently produce high quality results for this problem. Regarding the total risk of the UAV crashing as the objective, a discrete routing model is established. Based on a nonlinear relationship between grid areas and risk, an evaluation optimization model for the UAV is also established, and the value of potential areas is introduced to improve it. The simulation experiments verify that the two algorithms can both reduce the operation time and find the target in less risky situations. Results indicate that the Greedy algorithm is robust, and it exponentially drives toward high-quality solutions in relatively short time. While the Q-learning algorithm prefer to get less risky solution.