Cognitive decision making for navigation assistance based on intent recognition

Abstract

Within rehabilitation robotics, machines are being designed to help human in activities of everyday life. Mobility is an essential component for independent living. Autonomous machines with their high degree of mobility are becoming an integral part of assistive devices leading to a number of developments in mobility assistance. This is primarily in terms of smart wheelchairs embodied with agents. Autonomous agents keep an eye on irregularities during navigation and trigger corrections whenever required. They behave as teammates for the human wheelchair user. Such agents will be more effective if it’s behavior is closer to human or it is intelligent enough to understand the possible course of action taken by the human user. Therefore recognizing intention of the human driver and surrounding vehicles is an essential task. We have formulated a fuzzy model for the prediction of intention. A qualitative distance and orientation mechanism have been adopted, where few environment features are taken to show how the prediction of intention can improve the ability of decision making.