Research On Monocular Measurement And Target Tracking Of Wheeled Mobile Robot Based On Monocular Vision

With the improvement of science and technology of China, wheeled mobile robot(WMR) has been ap-plied in the fields of industry, agriculture, service, national defence, space exploration, and so on. These robots have a positive impact on the production and life of human society. Fruitful results have been achieved in research of WMR. Owing to the complexity of the environment and the diversity of tasks, autonomy and intelligence of wheeled mobile robots need further research. This paper discusses the research of monoc-ular measurement and target tracking of MT-R WMR based on monocular vision. The main contents are summarized as follows:(1) In order to solving the problem of target recognition in target tracking task of MT-R WMR, this paper discusses a kind of target recognition based on OpenCV. By using shape feature of target, a edge detection method based on target's outline is proposed to obtain candidate targets. Then, combining the method of threshold segmentation based on target's color feature, location of target in the plane can be achieved by cross matching of both features. Experimental results show that the recognition system can identify the target successfully.(2) Considering the need of real-time measurement in target tracking task of MT-R WMR, this paper propose a new method to measure the target distance for WMR based on the monocular vision, which over-come the deviation brought by feature points matching and optical axis deviation. This method extend the plane objective to the three-dimensional one and achieve high measurement accuracy without MATLAB ad-justments. First, by establishing the pinhole imaging model, this paper sets up the corresponding relation between world coordinate system and pixel coordinate system. Then, with respect to the singularity caused by matrix transformation, the concept of area is introduced and the relation between target distance and size of pixel area is achieved under a specific singularity case. At last, verification experiments are carried out on the MT-R WMR. Experimental results indicate that the comprehensive error ratios of the system are all under 0.7%, which satisfies the system requirements of real time, feasibility and effectiveness for monocular measurement of WMR.(3) The visual servoing control of WMR based on point stabilization is investigated. Firstly, by estab-lishing kinematic model of WMR and virtual tracking target, a kind of controller is proposed by combining position and attitude difference between target and WMR. Stabilization of the closed-loop system is analyzed by Lyapunov's stability theory and LaSalle's invariance theorem. Then, an improved controller is proposed to solve the problem that the forward speed of former controller is affected by angle errors. Stability of new closed-loop system is also proved by Lyapunov's stability theory and LaSalle's invariance theorem. Finally, numerical simulation and robot experiment both indicate that the improved controller can control WMR to track the target quickly and effectively.