Research On Obstacle Avoidance Of Guest Robot Based On Multi-sensors Information Fusion

With the development of robotics and artificial intelligence,intellectuality has been one of the important themes of the robotics technique. Self-obstruction problems must be solved in order to achieve intellectuality for the guest robot.Because of the complexity and unpredictability of the environment around guest robots, environment information can hardly be expressed in detail by single sensor.So the application of multi-sensor information fusion technology is inevitable, which is one of the key technologies to solve the obstacle avoidance problem. However, the autonomous obstacle avoidance problem has not received enough attention in the process of guest robot development. Therefore, the guest robot is taken as the platform and the application of multi-sensor information fusion in the obstacle avoidance of the robot is researched, which is beneficial to the application and popularization of the robot.In this thesis, the fuzzy logic information fusion and fuzzy neural network information fusion application in the obstacle avoidance problem of the guest robot are researched. And the fuzzy neural network algorithm of traditional T-S model is simplified. So the neural network training time is decreased. Then the availabilities of the two algorithms in solving obstacle avoidance problem of the guest robot areverified.Firstly, the application of multi-sensor information fusion technology in robot is studied and the fusion scheme of robot's motion environment information is proposed by using ultrasonic sensor, infrared sensor, electronic compass and binocular camera. Meanwhile the robot's motion model is established on the basis of the kinematic analysis.Secondly, the fuzzy controller is designed for the guest robot to avoid obstacles by researching fuzzy logic information fusion. Based on the MATLAB platform, the simulation experiment of robot fuzzy control and the fuzzy neural network controller are designed. Through the experiment, it is found that the guest robot under the fuzzy control can recognize the environment to realize obstacle avoidance. But when the environment is complicated, many problems will appear such as control rules match confusion and left and right swing, which is not conducive to the safe operation of the robot.Then, the T-S model of fuzzy neural network information fusion is studied and the simulation experiment is designed. It is found that the obstacle avoidance stability of the guest robot is improved, the response of obstacle avoidance action is more rapid and the problem of the fuzzy control is effectively solved.Finally, the improved fuzzy neural network algorithm is transplanted to the single-chip computer and the guest robot computer control program is developed.Through the obstacle avoidance experiment,the problem that robot swings can be dealt with effectively by the improved fuzzy neural network algorithm in this thesis and at the same time the stability of obstacle avoidance can be highly improved.