• A constrained multi-objective framework (DLCMO) is presented. • Auxiliary tasks gradually move from global to local searches. • Gradual shift of auxiliary tasks from objective to constraint information. • DLCMO has strong feasibility, exploitation and exploration capabilities. • DLCMO demonstrates competitive performance in UAV path planning problems. While numerous constrained multi-objective optimization algorithms have been proposed, effectively balancing convergence, diversity, and feasibility remains challenging in the context of unmanned aerial vehicle (UAV) path planning within complex environments. To address this issue, this paper introduces a constrained multi-objective optimization algorithm based on dynamic auxiliary tasks (DLCMO), which incorporates a dynamic selection preference and a dynamic global–local constraint assistance mechanism. In this approach, auxiliary tasks provide differentiated information to the main task throughout the evolutionary process, thereby synergistically improving overall performance. Specifically, the dynamic global-local constraint assistance (CAP) utilizes an ε-constraint method to promote uniform exploration of the potential solution space, effectively enhancing the diversity of the main population. Meanwhile, the dynamic selection preference assistance (DAP) adjusts selection preferences adaptively, which not only helps the main population traverse large infeasible regions but also significantly enhance convergence performance. Comparative experiments with seven state-of-the-art algorithms demonstrate that DLCMO achieves superior or competitive performance across 33 benchmark test functions and three UAV path planning problems, confirming its strong potential for real-world UAV applications.
Zheng et al. (Sun,) studied this question.