Path planning self-learning Algorithm for a dynamic changing environment

Abstract

Safe and optimal path planning in a cluttered changing environment for agents’ movement is an area of research, which needs further investigations. The existing methods are able to generated secure trajectories, but they are not efficient enough to learn from their mistakes, especially when dynamics of the environment are concerned. This paper presents an advanced version of the Ant-Air algorithm, which can detect the changed scenario and while keeping the lessons learnt from the previously planned safe trajectory, it then generates a safe and optimal path by avoiding collisions with the obstacles. The method presented can learn from the experience and hence improve the already generated trajectories further by using the lessons learned from the experience. The concept developed is applicable in various domains such as path planning for mobile robot, industrial robots, and simulation of part movement in narrow passages.