Formulation of a lightweight hybrid AI algorithm towards self-learning autonomous systems

Abstract

Autonomous systems able to react and change their behaviour in response to events during operation. These established abilities are based on the preprogrammed action or actions to be taken when encounter certain states in the deployed environment. Therefore prior to deployment, a knowledge expert with comprehensive understanding of the physical system and the deployed environment must specify anticipated states and determine actions to be taken to be programmed. As an alternative, a system with self-learning capabilities allows the system to autonomously identify, differentiate and classify states and progressively determine actions. In this paper, we present, summarize and discuss the formulation of a hybrid AI algorithm which combines Q-learning and AUTOWiSARD algorithm that will allow an autonomous system to self-learn. The hybrid AI algorithm will be implemented in an autonomous mobile robot simulation and the outcome will be presented and discussed.