Research On Navigation Of Intelligent Car Based On Monocular Vision Feedback

The research of intelligent transportation system has been paid a lot of attentions by many people at home and abroad and made some achievements. The research of intelligent vehicle direction based on the visual navigation has been a research focus due to the reduction of the occurrence of traffic accidents. The main researches of visual navigation include road detection, location of vehicle and control of it. The detection of road provides the justification of visual navigation. The difficulty of vehicle location is how to get the pose of vehicle according to a road image. Driving control of vehicle control has been a mature technology. But researches of lane-changing control are rare.According to three aspects proposed, the main research work in this paper includes:1. The information of road videos is complicated and incomplete, therefore there exists outstanding contradictions between fast classification and recognition. The real-time and validity of road detection algorithm are hard to ensure. The viewpoint of evolutionary vision process has been proposed. The road surface and lane will be detected by the process from scratch and information fusion of time and space. The lanes will be separated by a vanish point.2. According to the bounded ness of conventional camera calibration, the paper builds a binary quadric mathematic model to the relation of vehicle pose and image formation, and verifies the validity of the model through the significance of regression equation. The relative pose between vehicle and road can be directly got from road image by using the model which can locate the vehicle without the camera calibration.3. The research of lane-changing control based on visual feedback is rare, but changing of lane when the road is crowded or barriers appear is an important act which ensures safety running of the bodywork. With regard to lane-changing control, the paper proposes the process of describing the lane changing process to piecewise control process. The position error and angle error will be the input of the controller, and the lane-changing control can be realized ultimately.At last, the research work is summarized and the following working thoughts are proposed which will be of some guidance significance to the further research.