Abstract: To solve the optimization problem of 2D and 3D path planning in complex environments, we propose a path planning method based on the improved crested porcupine optimization (ICPO) algorithm. Firstly, we improve the standard CPO algorithm by combining an elite guidance strategy, an improved dynamic population adjustment strategy, and a parameter adaptation mechanism, which effectively enhances the global search capability and convergence speed of the algorithm. Secondly, we use the grid method to construct a 2D operating environment for the robot, and establish a comprehensive fitness function incorporating path length, safety clearance, and smoothness. Thirdly, we conduct simulation experiments in 2D scenarios containing multiple obstacle types. The results demonstrate that the optimal path length obtained by the ICPO algorithm is reduced by 3.26% to 5.99%, the average number of iterations is decreased by 16 to 80. It also effectively avoids various obstacles. Fourthly, we employ the Gaussian function to simulate a 3D operating environment for the robot, and establish a comprehensive fitness function integrating path length, safety, terrain fit, and smoothness. Finally, we perform simulation experiments in 3D scenarios with multiple obstacle types. The results indicate that the average fitness performance of the ICPO algorithm is improved by 1.66% to 20.97% compared with three other algorithms, thereby validating the effectiveness and feasibility of the proposed algorithm in path planning.