Perception in Remote Navigation

Abstract

This dissertation focus on the enhancement of human-robot collaborated remote navigation performance, specifically from the perception perspective. Three independent factors in the navigation-related perception were studied respectively, and enhancements for each of the factors were proposed, implemented and evaluated.\nTo study the nature of remote navigation, an extensive analysis of nature of navigation as well as difficulties in remote navigation were conducted. Basing on the understanding of remote navigation, scale perception (perception of the controlled vehicle's size relative to other object's size), distance perception (perception of distance between two objects), and orientation awareness (the awareness of current location, moving direction, and object's location relative to current location) were identified as critical factors. For the distance and scale perception, two experiments were designed specifically for each of them to quantify the remote perception condition's impact. It was found that under a remote perception condition, where human operator was separated from the environment where the navigation took place, both distance perception and scale perception were significantly impaired as compared with that obtained under a direct perception condition. Moreover, robust ways to enhance each of these three factors were also proposed and scientifically verified.This study successfully proved that with each specifically designed enhancement, the corresponding factor was significantly improved.\nResult of this dissertation work can be applied to various human-robot collaborated applications, such as urban search and rescue. With an integration of the three enhancement, human operators will have less failure in driving robot through hallways, doors or getting around obstacles, as well as have a more accurate understanding of the area's layout when a map is not available.