Research Of Reasoning Mechanism For Intelligent Robot Based On Multagent-Agent Architecture

CBR (Case-Based Reasoning) is an important artificial intelligence method, usingsimilarity matching method, which retrieves the most similar case with the new case from thecase base, then modifies the case base to provide a solution for the new case. As the CBR couldimitate human’s cognition process well, we can apply it to the field of intelligent robots to makerobots more intelligent. At present, CBR reasoning mechanism has become a hot topic inintelligent robots. In addition, as the Agent-oriented programming (Agent OrientedProgramming, AOP) proposed, while the Agent component has self-control capability, canmake active decision for incident response in real-time environment, software design based onAgent and Multi-Agent System (MAS) has significant meaning and value in the field ofintelligent robots. Multi-Agent System requires a “brain” of reasoning, solving, and learning topropose solutions for the problems encountered, and CBR can solve this problem well, sointelligent robot system MARS (Multi-Agent robot system) combining CBR and MAS is agood research topic.In MARS, the delay of CBR reasoning mechanism itself has a great impact for thereal-time of system, so improvement must be done to better apply the CBR reasoning engine tothe MARS. This paper firstly describes the knowledge of the CBR reasoning mechanism andthe reasoning algorithm of various stages, and puts forwards a distributed single-robotMulti-Agent Architecture, mainly to research and design CBR reasoning mechanism indeliberating Agent. Then it analyzes the real-time of CBR and makes improvement by usingspace vector method, combined with dynamic-weight assignment, the optimal cost, andtime-bound algorithm.Finally, an intelligent robot experimental platform is built using modular components. Aftertransplanting the complete software system and improved algorithm to actual intelligent robotplatform, we achieve the actual test data, verify, test, analyze and evaluate the structure of thissystem and the performance of improved algorithm.