Path Post-Optimization Based on Search Space Adjustment and CMA-ES for Microrobots Navigation in Complex Environments

Magnetic field-driven microrobots have exhibited great potential in the medical field, wherein navigation safety in the complex human body environment depends on path smoothness, computational efficiency and obstacle avoidance performance. However, existing path planning methods perform poorly on one or more aspects. In this paper, an efficient post-optimization method that is insensitive to the quality of the original path and adaptable to complex environments to gain a high-quality path is proposed. With path stretching and appropriate path discretization for the search space adjustment (SSA), the Covariance Matrix Adaptation Evolution Strategy (CMA-ES) manages to meticulously explore the local space and quickly find a workable path meeting strict requirements. In the simulation, compared with the original path generated by RRT*, the optimized paths are significantly improved in terms of length, smoothness, obstacle avoidance performance with relatively high computational efficiency.