Research Of Vision Navigation System For Wheeled Mobile Robot

With the development of control technology, intelligent robot has become a hotspot in the field of automatic control. Taking the robot visual navigation and control issues as the subject, a research on the design of control system, information accessing, processing and identification of objective road is addressed in this thesis. And based on these, effective control algorithms are used to control the robot so that the robot can track the road accurately and quickly.Taking the 32 STM32F103 as the core and adopting a CCD camera, the hardware system is constructed based on the functional analysis of the robot. Furthermore, the design of the core circuit, power module, serial module, speed check module, line module and motor drive module is accomplished in the thesis.In the thesis the embedded system is set up on the basis ofμC/OS-Ⅱoperating system, which resolves the real-time problems in the traditional single-mandate system, and enhances the reliability of the robot system through real-time multi-task scheduling management. With the actual mission, theμC/OS-Ⅱtransplant in STM32F103, and the specific mandate of the scheduling process underμC/OS-Ⅱare described.After researching the robot vision navigation algorithms, a non-uniform collection and coordinates transformation method is proposed to achieve the camera calibration, which is applicable to the specific environment and scarce computing resources. Furthermore, a progressive dynamic threshold calculation is adopted to identify the path. The experiment results prove that it is effective and has good adaptability in uneven light environment.A control strategy of robot system is proposed and a motion model is established based on speed difference, speed and turning angle of the steering gear in the rear wheel. Moreover, the characteristics of the system model are analyzed. A steering gear control strategy based on classic PID algorithms and fuzzy control algorithms is investigated in the thesis, and the rear-wheel-drive motor is controlled according to the speed information obtained through speed sensor. Moreover the experimental results prove that the proposed method is effective and advanced.