Implementation of machine learning approach by the application of ant colony optimization with enhanced heuristic within the realm of robotic path planning

Abstract

The current study delves into the utilization of Ant Colony Optimization with enhanced heuristics within the realm of Robotic Path Planning. Traditional ACO algorithms often exhibit slow convergence and a moderately high likelihood of getting stuck in a local optimum, issues that this algorithm aims to address. The algorithm is simulated in two environments: MATLAB and ROS (Robot Operating System). The heuristics are adjusted, and the resulting convergence and path lengths are meticulously recorded and analyzed. Subsequently, this algorithm is integrated into the global path planner package in ROS, a component of the robot’s navigation stack. The robot is then successfully simulated to navigate from its initial position to a specified goal location. The results showed a runtime of about 72% faster, a path length within the acceptable range, and a shorter turn angle. The algorithm is more practical and capable of producing a smoother navigation path suitable for real-time robot path implementation.