Information Fusion Technology In The Home Service Robot Navigation

This thesis concentrates on the study of the robot's intelligence information fusion and automatic navigation in the home environment that is occasionally changing with uncertainty. The related artificial intelligence technologies such as blackboard model and neural network are integrated into the information fusion methods to implement the intelligence fusion. The corresponding works are summarized as follows:(1) Considering the specialty of home service robot, a distributed fusion architecture based on blackboard system is presented, and also an integrated intelligent home service robot navigation system based on it is designed. It takes the advantage of distributed fusion method which can flexibly and rapidly dispose information, and expert knowledge based on blackboard model is used conveniently to improve the intelligence level of information disposal.(2) Since there are unavoidable measure errors of sensors, a home environment recognition technology based on dynamic neural network is proposed, and it is used to fuse the spatio-temporal information gathered from sonar and PSD sensors. According to this technology, robot can recognize and understand the home environment intelligently when moving around. This method is used to achieve dynamic environment information fusion, and then the robot can get timely information when navigating.(3) The information in both of the feature and decision layers is fused together, and a hybrid information fusion method is presented to carry out robot's local path planning. The CCD and laser sensor information is fused to detect environment information and path conditions in the feature layer, then in the decision layer D-S evidential theory is used to improve the confidence level of information and a better path is guaranteed to be found.(4) The emotion-based behavior coordination system is designed and realized. The robot can imitate the human's emotion and decision-making mode to adapt to the dynamic environment. Moreover, the local path planning technique based on the landmark information fusion is integrated into emotion-based behavior coordination to control robot's movement and navigate the robot. Simulation result shows that by means of this method the robot can navigate more rapidly and selects better route.