Based on the limitations of traditional pigeon swarm optimization algorithms in dealing with high-dimensional optimization problems， a modified spiral-dynamic pigeon-inspired optimization （MSDPIO） algorithm is proposed to solve problems such as slow convergence speed and susceptibility to local optima in robot path planning. Firstly， by introducing the logistic chaotic mapping initialization strategy， the search range is expanded. Secondly， spiral search strategies and dynamic reverse learning strategies are designed to improve the position update mechanism and enhance the convergence speed and resolution quality of the algorithm. At the same time， the adaptive cosine function is used to adjust the reverse learning weights and improve landmark operations to enhance the algorithm’s adaptive and global search capabilities. The performance of the algorithm is effectively evaluated through experiments on ten CEC2017 benchmark test functions. MSDPIO algorithm and improved B-spline curves is applied to path planning problems on maps of different scales （20 m × 20 m and 40 m × 40 m）. Simulation results show that in a small-scale 20 m × 20 m map，MSDPIO algorithm improves the path length by 1.46%， 1.43%， and 1.47% compared to pigeon-inspired optimization （PIO） algorithm， multi-strategy improved aquila optimizer （MSIAO） algorithm， and grey wolf optimizer （GWO） algorithm， respectively. The maximum convergence performance improvement in a large-scale 40 m × 40 m map is 37.82%.