Research And Design Of Dm6437-Based Vision Systems For Mobile Robot

Mobile robot vision refers to the mobile robot with vision sensors which can let it has the ability to identify, detect and decide autonomously in unknown environment. It shows great potential in the assembly process, intelligent household, nuclear accident treatment, and modern war investigation. Therefore, the research of mobile robot vision system has become one of the important subject for the scholars.Compared with the traditional machine vision, vision system of mobile robot appears some new problems, which mainly contain the following aspects. Firstly, the traditional vision system that is made of the image acquisition card and computer has high power cost. To increase the endurance time of the mobile robot, the low cost system must be adopted. Additional, under the influence of light, partial occlusion and vibration, how to improve the accuracy and efficiency of the feature matching, is also a key issue for the mobile robot pose estimation. Last but most important, in the present the technologies of the mobile robot, the major difficulty is how to simplify the pose estimation algorithm and avoid singular problem. According to the above problem, this thesis mainly aim to study vision system of the mobile robot from the following three aspects.According to signal integrity simulation and analysis, a high-speed digital system based on DM6437is designed to meet the requirement of high performance, low power consumption and compact structure. The thesis discusses systematically the generation mechanism of reflection and crosstalk, and accomplishes the modeling and simulation of the key transmission line by the SigXP and the constraint manager (CM) tools in Cadence. According to the simulation results, some methods for suppressing the reflection and the crosstalk, reasonable topology and traces are given. At last, the experimental results show that the hardware of the system is stable. To reduce the effect caused by the noise, partial occlusions and robot vibration, image features matching algorithm is designed for position and pose estimation of mobile robot. Firstly, based on the features obtained by integrating FAST-9and SAD methods, the16dimension descriptors are established. And then features are matched by the nearest neighbor method. Finally the false matching points are eliminated by random sample consensus algorithm. The experimental results show the effectiveness of the method.Considering the motion constrain in two dimension plane, the fast pose estimation algorithm based homography is adopted in this work. The experimental results show that the present method can avoid the singular value and effectively obtain the position and pose of the mobile robot.