Highway Autonomous Driving Vehicle Road Testing Technology Research

This dissertation has researched on Lane Detection technology for AHV(Autonomous Highway Vehicle), implemented the inverse perspective projection from perspective road image to world coordinates and proposed Lane Detection arithmetic for projected image.A simplified camera projection model has been researched. This projection maps the perspective road image into projected image and removes intense perspective effect in the former one. In projected image, the structure character of road is highlighted.A low-level process has been proposed according to the structure character of Lane Marking. Guided by this character and quality character of projected image, a Lane Marking edge detection operator has been designed, and a method for Lane Marking pixels enhancing and multi-segment thresholding has been proposed.A novel Lane Marking search arithmetic has been proposed based on the polynomial lane model which represents the structure character of lane. Lane Marking can be detected by three steps of this arithmetic: firstly calculating lane width with statistical method, then tracking Lane Marking with search model method, finally correcting the track result with lane model curve-fit method which improves reliability of detection results.As verification, this arithmetic has been tested by simulating experiment which has been especially built, and the results are introduced.