Research Of Mobile Robot Path Planning Based On Stereo-vision

Stepping into the 21st century,robot are applied into industry production and our lives more and more widely,it has also represented science and technology development level of a national.So the research on robot is of great significant.Machine vision system with characteristics such as non-contact measurement,large signal range and complete information plays an important role in the development of robot technology.Based on stereo vision,this paper do some studies on its application in robot path planning.In this paper,efforts focus on research of stereo vision system,camera motion analysis,clustering algorithm in3D reconstruction image application and mobile target tracking.Two CCD cameras were used to make up a stereo vision system,finished the work of single camera calibration and stereo calibration,reconstructed the 3-D environment.A method of quantitatively estimating camera motion parameters used in the vision-based mobile robot navigation is presented,using the feature points' image coordinates between two adjacent frame to the projection transformation model to get the camera movement parameters.The FCM algorithm uses only Euclidean metric in measuring distance between vectors,so it has not ideal performance when utilized in high dimensional data,based on FCM this paper introduce Genetic algorithm and the kernel method for distance measurement.The improved algorithm is applied to segmentation of the 3-D reconstruction images,experiments show that the improved FCM algorithm can effectively avoid the interference of noise on image.Finally,according to the defect of CamShift algorithm,we introduce Kalman filter for improving.In each frame interval,with the predict results of the Kalman filter as the initial position of the CamShift search box.This improved algorithm can effectively avoid failure caused by moving obstacle shielded.