Reinforcement Learning in Simultaneous Localization and Mapping
Springer Nature (ISBN: 978-981-96-0047-2). Read the chapter
Abstract: The primary objective of this research work is to investigate model-free path planning for reconfigurable robots using value and policy iterations. The focus is on developing and evaluating an autonomous algorithm for robot path planning. Initially, the A-Star algorithm was modified to incorporate model-based learning. Subsequently, model-free reinforced learning was incorporated through value iteration and policy iteration. The experimental validations of modified A-Star algorithm is conducted in python virtual environment.