Real-time camera based lane detection for autonomous driving

In this report a new designed real-time lane detection algorithm based on monocular vision and inertial measurement unit (IMU) is explained and evaluated. The acquired image from the camera is undergoing an inverse perspective mapping to create a bird-view perspective of the road. Local maxima in this image determine the lane-markers. Faulty local maxima are removed and constrained line fitting over these local maxima is applied using Quadratic Programming. This fit represents the current drive-lane existing out of the left and right lane marker. The lane-width and vehicle distance to the center of the lane are calculated with the use of these fits. The vehicles IMU is used to predict the lane-marker in the next time-frame. The result is a lane detection algorithm which is working in real-time at a frequency of 10Hz. The algorithm is validated and is able to successfully detect the lane-width and vehicle distance to the center up to 94% of the time in highway scenarios. For almost every occurrence where the algorithm performed poorly, an error signal is send. In most of these situations the algorithm is able to minimize the error via logical statements and rely on previous measurements. It can be concluded that this algorithm can be used to maintain the vehicle position inside the lane for highway conditions. Real-time camera based lane detection for autonomous driving iii