| Nowadays,the complexity of robot behaviors grows,and static operating environment becomes more dynamic and open.The autonomy of robot behaviors and the safety of software challenges conventional robot control software in various aspects,such as automated planning,concurrent tasks and real-time interaction with environment.How to fast and efficiently develop such software remains to be solved.This paper develops an agent-oriented framework to support development of autonomous robot control software and proposes a behaviors detecting method that enables robot to detect its behaviors’ safety before their executions.The framework is structured in hierarchical layers,which contains decision,functional agent and execution layers.There are three kinds of safety issues considered,including hardware constraints,stability and collision confliction.The corresponding algorithm ACD algorithm for collision confliction detecting are designed.Implementation framework for the method technology is proposed which is based on AOP.Experiments have been conducted on the humanoid robot NAO,results of which prove such framework to be highly efficient,low coupling,reusable and well structured,and the method can effectively detect harmful behaviors and adjust motions in compensation for behaviors. |
