Service Robot Navigation Based on Q-Learning and Fuzzy Logic

Abstract

In this paper, a new approach is developed for solving the problem of service mobile robot navigation in a semi-structural dynamic environment based on Q-learning and Fuzzy logic. Service robots usually work in an indoor environment like a nurse robot or reception robot. Indoor environment is a semi-structural, because the map of building is known but some obstacles like humans also are existed in the robot path. In this paper Navigation is divided in two parts: known state and unknown state. In the known state, there is not any dynamics or statics obstacles accepted walls of building, in this state robot uses fuzzy logic to reach a goal. In unknown state, there are every obstacle and wall, in this state obstacle avoidance is done by Q-learning algorithm.