Intelligent decision-making strategy for multiple humanoid robots path planning in complex environments using an Improved Adaptive Artificial Fish Swarm Algorithm
TL;DR
An Improved Adaptive Artificial Fish Swarm Algorithm for planning the navigation paths of humanoid robots in environments containing both static and dynamic obstacles and consistently outperforms conventional path-planning techniques in terms of both path length and navigation time in unknown environments.
Abstract
Swarm intelligence-based path planning techniques have been widely implemented for solving robot path planning problems due to their effectiveness in unstructured and complex terrains. However, many existing swarm intelligence techniques are robust, they suffer from low convergence speed, premature stagnation, low optimization accuracy, and often tend to get trapped in a local optimum point. To overcome these limitations, this paper proposes an Improved Adaptive Artificial Fish Swarm Algorithm (IAAFSA) for planning the navigation paths of humanoid robots in environments containing both static and dynamic obstacles. The proposed method incorporates several enhancements, including chaotic population initialization using a logistic map, dynamic visual range and step-size adjustment, improved swarming and following mechanisms with adaptive step scaling, and an adaptive elimination–regeneration strategy based on the t -distribution mutation mechanism to enhance population diversity and convergence performance of the traditional AFSA. The proposed approach was tested in simulation and real-time trials using a physical NAO robot. The experimental findings showed that the IAAFSA produced shorter and smoother paths while maintaining collision-free navigation. The IAAFSA framework is combined with a Petri-Net controller to coordinate interactions among multiple robots and prevent deadlock during navigation. The outcomes reveal that the deviation is maintained at less than 5% for single-robot contexts and under 6% for multi-robot navigation. The reliability and robustness of the developed navigation strategy for humanoid robot path planning in complex environments are confirmed by the strong agreement between simulation results and real-time testing findings. The proposed method consistently outperforms conventional path-planning techniques in terms of both path length and navigation time in unknown environments.