Abstract: Objective: Path planning for unmanned aerial vehicles (UAVs) in complex mountainous environments remains a significant challenge due to steep terrain, limited navigable airspace, and multiple overlapping constraints. Conventional metaheuristic algorithms, including Particle Swarm Optimization (PSO), Grey Wolf Optimizer (GWO), and Harris Hawks Optimization (HHO), often suffer from premature convergence and unstable performance when addressing such nonlinear, high-dimensional problems. To enhance global search capability, convergence stability, and optimization precision, this study proposes an improved metaheuristic algorithm named Crested Porcupine Optimization with Chaotic Mapping and Hybrid Mutation (CPO-CH). The objective is to achieve efficient and safe UAV flight path planning under multiple terrain and flight constraints. Methods: A multi-objective cost function was constructed to comprehensively consider flight constraints, including terrain obstacles, no-fly zones, altitude limits, path smoothness, and safety distance. Based on the original Crested Porcupine Optimization (CPO) algorithm, several key improvements were introduced. First, a chaotic mapping initialization method was adopted to generate a diverse and uniformly distributed initial population,thereby increasing search space coverage and avoiding early stagnation. Second, a hybrid swarm collaboration mechanism integrating the cooperative learning principle of PSO with the defensive behavioral strategy of CPO was designed to achieve dynamic equilibrium between global exploration and local exploitation. This mechanism enables adaptive reverse learning between optimal and inferior solutions, improving exploration ability in complex constraint spaces. Third, a Cauchy–Gaussian hybrid mutation mechanism was developed to balance coarse-grained global search and fine local refinement. Furthermore, a cyclic population reduction strategy was employed to accelerate convergence while preserving population diversity during later iterations. The proposed algorithm was applied to UAV path planning in a three-dimensional mountainous terrain model constructed from real Digital Elevation Model (DEM) data. Results: The proposed CPO-CH algorithm achieved a minimum path cost of 605.4973, which is 1.7% lower than that of GWO (616.0432), 16.0% lower than that of the original CPO (720.9759), and 27.3% lower than that of HHO (832.9132). Although requiring moderately longer computation time, CPO-CH achieved the best balance between precision and stability. Benchmark tests on CEC2005 functions confirmed that CPO-CH obtained the lowest mean and standard deviation values, indicating superior convergence robustness and global search capability. Conclusions: The CPO-CH algorithm provides an effective and quantitative solution for UAV path planning in complex mountainous environments. The integration of chaotic mapping, swarm cooperation, and hybrid mutation significantly improves global exploration, convergence efficiency, and path quality. This approach demonstrates strong applicability for safe, energy-efficient, and smooth UAV operations. Future research will focus on adaptive parameter adjustment and multi-UAV cooperative optimization to further enhance real-time adaptability and scalability.