An Integrated Experience-Based Approach to Navigational Path Planning for Autonomous Mobile Robots

Navigationalpath planning is a classicalproblem in autonomous mobile robotics. Most AI approachesto path planning use goal-directedheuristicsearch of problem spaces defined by spatial modelsof the navigationspace. Thispaper exploresan alternativeapproach that integratesa new experience-based method with the traditional model-based method. Core issues in using experience-basedmethods include the content, representation and indexing of past cases, organizationof the case memory, retrieval of casesfrom memory relevant to the current problem, adaptation of retrieved cases to meet the specificationof the currentproblem, and verificationof the adapted solution to the problem. Our hypothesis is that spatial models of navigation spacescanprovide answersto some of theseissuesin experience-basedpathplanning. TheRoutersystem m i n e s thishypothesisin thecontext of path planning in geographical spaces. It uses a hierarchically-organizedspatial model of the navigationspace to index the casesand t...

• Type: Conference
• Year: 1993
• Where: ICRA
• Authors: Ashok K. Goel, Michael W. Donnellan, Nancy Vazquez, Todd J. Callantine
• Keywords: AI Approachesto Path, Case Memory, ICRA 1993, Path Planning, Robotics