| It is one of the research hotspots to improve the intelligence of mobile robot. With the development of the intelligence, adaptability, rapidity, accuracy and good reliability are demanded more in changing environments. It is studied in this thesis how to control the mobile robot in changing environments intelligently, and an experimental platform is built up to verify the method. The main work and conclusions are shown as following:(1) A low cost and universal experimental platform is built up whose core sensor is a camera. It is shown in practical use that the platform meets the purpose of verifying the adaptive method based on machine learning when mobile robots in changing environments.(2) The classification accuracy of SVM which is effected by the different lane features is studied, and the LIBSVM is used to try out. The results show that using the lane of center line can get higher accuracy.(3) A method is proposed to employ SVM in the limited hardware resources of microcontroller. The time complexity of this method is analyzed. The run time of the experiments shows that models with a dozen of features and dozens of support vectors running in many microcontrollers could meet the real-time need for most mechatronic systems.(4) A steering model is established to calculate a new streering function which possesses less parameter variables, smaller coupling and more precise steering angle. It is a good way to solve the problem of the inexact steering and delay in changing environments. The simulation results show that this method can meet the requirements of accuracy and the adaptability of mobile robot in different lane types.(5) A fuzzy PI controller is used to solve the problem of response slowly and liable of the mobile robot in changing environments. Simulation and experimental results show that the controller can adapt to the changing of speed quickly when switch to the various types of lanes. It can also improve the stability of mobile robot. |
PDF Full Text Request