Mobile Robot's Wall Following Based On Neural Network And Fuzzy Control

At Present time, more and more various kinds of mobile robots are being introduced into many application situations, such as automatic transfer of articles, automatic exploration of dangerous regions, and service work etc. High demands on the navigation ability of mobile robot are put forward by these application, the ability of road recognition and road following is an important sign of the mobile robot抯 intelligence level among many aspects of navigation technology. But real-time and static control algorithm as well as the control strategy limits the recognition ability along with the road following velocity and precision. For this reason, this dissertation presents a systematic study on the road recognition and road following theory and strategy of mobile robot. In this dissertation, the main work includes: (1) The system structure and control principle of TIT-I mobile robot is analyzed and introduced. (2) Fuzzy control and neural network can improve real-time control performance for mobile robot due to its high robustness and error-tolerance ability. In this dissertation, after the basic principle of fuzzy control is introduced, the shortcoming of the traditional method for looking up the fuzzy control table is pointed out and an improved method based on linear interpolating is put forward for the first time. Antitheses experiments show the great improvement, what抯 more, the improved method is well interpreted on fuzzy control theory. Training a neural network in order to make it converge is an absolutely necessary step for using it, in this dissertation, the factors that affect the convergence of Bp neural network are studied by experiments, which can direct the training of it. (3) The ubiquity and comparative orderliness wall is used as the object of road recognition and road following for mobile robot. Only using the least 3 ultrasonic sensors, the mobile robot successively recognizes and steadily accurately follows the road by synthetically using neural network, fuzzy control and the idea of sensing-planning-acting. (4) According to the TIT4 mobile robot, A very actual simulation platform is made by adopting the idea of OOP and using the visual exploitation tool(VC++). In the end, the control strategy in this dissertation is verified on TIT-I mobile robot.