Sensory-data extraction based real time mobile robot motion planning using fuzzy logic

In this paper a real-time mobile robot motion planning system is proposed. This is done by extracting the sensory data of the IR sensors of the robot during offline training of the robot. The mobile agent is able to interact with an unknown environment using a reactive strategy determined by sensory data. After classifying the data into several classes we formulate the Gaussian membership function of each feature of the respective classes. The features here correspond to the sensory data of six IR sensors as well as the speed and alignment of the mobile agent with the predefined goal. After each iteration, the robot undergoes step movement to move to its immediate next sub-goal computing two control variables which are direction and speed. Thus it reaches its predefined goal. The precision of this work is proved by the fact that the robot's final destination through step movement ultimately coincides with the predefined goal. Moreover the performance of our approach for multiple mobile agents is found to have outperformed when compared to the classical heuristic and evolutionary techniques-based path planning strategies. The entire work has been implemented on Khepera II mobile robot.