Twin Robot Cooperation in Multi-Robot Environment: An Applied Q-learning

Abstract

This paper provides a new approach of executing the twin robot operation with the application of classical Q-learning and improved Q-learning algorithm. This approach has significantly less space and time requirement for execution of the algorithm As only the best Q-value with matched actions are stored for each state, the space requirement is reduced. Similarly, only the best matched state-action pairs with best Q-value are executed, it requires less amount of time for execution. On application of the classical and improved Q-learning in twin robot operation, the proposed algorithm takes the supremacy with respect to the space requirement and the traversal time.