A Decomposition-based Constrained Multi-objective Evolutionary Algorithm with An Infeasibility Utilization Mechanism for UAV Path Planning
Cai Zhang, Jinpeng Zhang, Lin Huang, Chaoda Peng · 2022 34th Chinese Control and Decision Conference (CCDC) · 2022
Evolutionary algorithm has been used to solve unmanned aerial vehicle (UAV) path planning problems. Usually, a UAV path planning problem is modelled as a constrained optimization problem, and the goal for evolutionary algorithms is to find a collision-free trajectory with respect to the constraints, which requires the algorithms equipping not only a powerful search engine but an effective constraint-handling technique. In the community of constrained multi-objective evolutionary algorithms, an essential idea is how to make good use of the informative infeasible solutions during the evolution process, which can significantly improve performance of the algorithms. To the best of our knowledge, this has seldom been explored in UAV path planning problems. To address this issue, this paper proposes a decomposition-based constrained multi-objective evolutionary algorithm with an infeasibility utilization mechanism for solving UAV path planning problem. UAV path planning represented by the B-Spline curve is first formulated as a bi-objective optimization problem, i.e., minimizing the travelling distance and the risk of a UAV, with three constraints including the minimum flight height, the maximum flight height and minimum flight angle. Then a decomposition-based constrained multi-objective evolutionary algorithm that can utilize the information containing in infeasible solutions is adopted to solve the constrained optimization problem. To further make good use of the infeasible solutions found by the algorithm, an infeasible utilization mechanism is proposed to guide the search to the optimal regions. The experimental results have indicated that the proposed algorithm is superior over the compared algorithm in terms of finding a set of well-distributed and well-converged non-dominated solutions.