Research On The Key Technology Of Mobile Robot Navigation System

With the advancement of science and technology, the rapid development of research-based visual robot navigation and positioning; visual technology is also known as an important area of research, advances in visual technology also makes it possible for the mobile robot has a broad application prospects. Current mobile robot is mostly based on the visual sensor; it will first collect the environmental information of the scene, and then the target identification, path planning, and ultimately complete navigation. In the entire process is crucial for vision technology. In this paper, mobile robot using binocular vision system, research in recent years has been made in technology, improved camera calibration method, target recognition, and image registration method, the hardware architecture, using the embedded system structure to move the robot design.Innovation in this paper and the main work are as follows:(1) Study camera model and calibration method, this paper using Tsai two-step method of camera calibration, this method is simple, and the method is a flexible, high positioning accuracy.(2) Improved corner feature matching algorithm to achieve a fast corner detection and matching, which rough matching and fine matching two-step process to achieve bidirectional grayscale calculated the corresponding point of crude matching using poplar constraint and the fundamental matrix achieve precise matching image, to obtain an exact match of the final image. In this paper, the proposed method has good robustness and accuracy, can guarantee the accuracy of image matching under the premise of providing real-time data to the mobile robot.(3) In this paper, a feature recognition method based on the same moment feature is used. The method for the pattern of rotation, translation and scaling invariance characteristics, by the feature extraction can be efficiently and accurately identify the target object.(4) The improved color lookup table for image segmentation and using particle morphology analysis to the area for target object recognition. Used improved path planning algorithm:the Viterbi algorithm path optimization, and reasoning the path, backtracking arithmetic in Matlab software simulation to achieve good experimental results.Image registration and image recognition, fast corner detection algorithm using improved feature extraction, after rough matching and fine matching two steps to obtain high-precision image matching. The experiments show that the two-step matching method can improve the matching accuracy and speed of the target. In this paper, the hardware design using ARM chip as the processor, FPGA chip to image fast calculation algorithm to improve the computing speed optimized image, so you can solve real-time problems in robot navigation.