Research On Obstacle Avoidance System Of Mobile Robot Based On Multisensor Information Fusion

Currently, the multi-sensor information fusion technology, that provided the theoretical support to further improve the intelligent level of robot, has been applied in the mobile robot progressively. Using the multi-sensor information fusion technology in the obstacle avoidance system of mobile robot, multiple sensor information can be coordinated to solve the nonlinear and uncertainty of robot obstacle avoidance system. It is of important meanings to improve the accuracy of the robot obstacle avoidance. Therefore, in the task, based on the toxic gas inspection robot which designed for the chemical industry of Shanghai, the application of multi-sensor information fusion technology in the mobile robot obstacle avoidance system is accomplished, and good results are obtained.In this paper, the obvious advantages of fuzzy logic control and fuzzy neural networks in nonlinear systems are found by studying the common multi-sensor information fusion algorithms. Therefore, we studied the multi-sensor information fusion algorithms applied in the mobile robot obstacle avoidance system, focus on the fuzzy logic control and fuzzy neural network.Firstly, the mobile robot obstacle avoidance system is designed by the kinematics analysis. Then, a fuzzy controller to avoid obstacles is designed by researching fuzzy control algorithm and determining the input and output membership functions and fuzzy control rules. With the Matlab simulation and physical verification, we proved the algorithm can meet the requirements of the mobile robot obstacle avoidance. But the control accuracy needs to be improved. To further enhance the adaptation ability of mobile robot in complex environment, the TS fuzzy neural networks which has self-learning ability is studied, and the initial membership function are trained off-line in the use of the networks parameter learning algorithm. The simulation and experiment shows that the fuzzy neural network for obstacle avoidance algorithm approximated the nonlinear systems more in high precision, and the reasoning ability of fuzzy control and the learning ability of neural networks are integration. Compared to fuzzy control algorithm, the fuzzy neural networks algorithm has higher accuracy and faster response .Based on the multi-sensor information fusion, the obstacle avoidance function of differential mobile robot is accomplished, which can guarantee the mobile robot works in more complex environments reliably.