Study On Obstacle Avoidance Of Mobile Robot Based On Multisensor Information Fusion

With the improvement of science and technology,sensor technology has been rapidly developed.Multisensor information fusion technology has been widely used both in military and non-military fields.Information fusion has become a prevalent research direction.In addition,the mobile robot technology is a key intelligent technology of the country and has been highly concerned by the world's scientific researchers.Obstacle avoidance technology is an important guarantee for mobile robot to complete the various tasks.In order to improve the accuracy of obstacle avoidance of robot in complexly unstructured environment and provide the corresponding theoretical basis,multisensor information fusion technology has been widely applied to mobile robot obstacle avoidance system.Through the artificial neural network,artificial potential field method and other methods which based on multisensor information fusion are applied in autonomous obstacle avoidance algorithm around the word,this thesis will focuses on the main idea of the multisensor information fusion technology.We analyzed the characteristics of each fusion algorithms and applications of mobile robotics.It is found that fuzzy logic and neural network are effective tools to solve the nonlinear problem of robot.Fuzzy control algorithm and T-S(Takagi-Sugeno)fuzzy neural network algorithm are discussed and studied respectively,and then found the advantages of combining fuzzy control with neural network by compared.Based on theory and practice,we explored the fusion algorithm in the application of robot obstacle avoidance.First ofall,we set up the mobile robot experiment platform base on the multisensor obstacle avoidance.According to the characteristics of the robot obstacle avoidance,we used ultrasonic sensors and infrared sensors to obtain distance information of the unknown environment obstacle and we alse can use the electronic compass to get the position and target azimuth of the robot.We preprocessed environmental information of each sensor,and then inputted to the fuzzy controller and fuzzy neural network controller respectively.By MATLAB simulation and hardware verificationto prove T-S fuzzy neural network algorithm is superior to the fuzzy control on the accuracy and response speed of the mobile robot in unknown environment.The effectiveness and feasibility of the proposed method in the mobile robot obstacle avoidance are verified in theory and practice.