<p>Currently, unmanned aerial vehicle (UAV) technology is widely employed across various industries owing to its inherent advantages. In terms of UAV technology, exploring and optimizing path planning for UAVs occupy a prominent research position. Thus, a constrained optimization model for the UAV path planning was developed, and then, the Marine predator algorithm (MPA) was applied to effectively solve this model. Nevertheless, the MPA encounters limitations, including the tendency to become trapped in local optima and suffer from premature convergence. Therefore, a modified version of MPA, which is called MMPA, was developed. Firstly, circle chaotic mapping is introduced into MPA to address non-uniform initial search agents’ distribution in the algorithm. Secondly, the neighborhood perturbation strategy is introduced to bolster MPA’s performance, enabling it to escape from local optima. Thirdly, in the later iterations of MPA, the lens-imaging-based learning strategy is implemented as a means to enrich search agents’ diversity and further improve the algorithm’s optimization capabilities. From the experimental reports, it is known that the performance of MMPA is better than that of the comparison algorithm, both in the benchmark functions and in UAV path planning. When it comes to path planning, the routes generated by MMPA are smoother and safer than those generated by the comparison algorithm.</p>