Based Multi-sensor Information Fusion For Mobile Robot Obstacle Avoidance

With the platform of AS-R mobile robot, the application of multi-layer fusion method for avioding obstacle for mobile robot was studied principally from the view of the multiple sensor information fusion.In this thesis, it was proposed that using a multi-sensor system for avoiding obstacle for mobile robot by using multiple ultrasonic sensors and infrared sensors to solve the environment information couldn't be precision and detail achieve by single sensor. By experiment, the performance of sensors was tested and the error of measuring diatance of sensors system has been presented.As the uncertain information of ultrasonic sensors, a two-step fusion method was proposed based on combining the BP neural network and fuzzy neural network.In this method, in order to realize inforemation fusion for multiple ultrasonic sensors, the BP neural network was used to the first step fusion algorithm from each sensor group to achieve a procision of obstacle identification; in order to realize decision control for avoiding obstacle for mobile robot, the fuzzy neural network was used to the second step fusion algorithm. The two-step fusion method achieved the performance of robust and flexible better than the information fusion of single layer. Moreover, the stratergy of avoiding obstacle was proposed and the avoid obstacle was realized for given environment for mobile robot. Finally, the simulation experiment and experimentlize with AS-R mobile robot verify that the above_mentioned mehod is really feasible and efficient.