A Behavior-based Control of an Object-pulling Robot Using Fuzzy Discrete Event System

Abstract

This paper describes a novel behavior-based approach for an object-pulling robot using Fuzzy Discrete Event System (FDES). The object-pulling task is a variant of objectpushing operation, where the robot anchors itself with an object and then navigates to a target location. The proposed behavior-based approach incorporates global and local motion planning for navigation. The global motion planning uses strategic behaviors that employ prior knowledge of the environment to generate a safe-path from the initial position of the robot to the end-point (goal), whereas the local motion planning uses reactive behaviors that deploy locally sensed sensory data for target-object detection and dynamic obstacle avoidance. The major contribution of this paper is the novel FDES-based behavior selection (or modulation) method that eliminates the conventional binary thresholding-based boolean event generation in Discrete Event System (DES) for behavior activation. The multi-valued logic approach of FDES reduces the possibility of wrong event generation, which in turn decreases the possibility of inappropriate behavior selection. Experimental results using a physical robot are also presented to authenticate the feasibility of the proposed FDES-based object-pulling operation.