LaneTrak: a vision-based automatic vehicle steering system

Publisher Website
Google Scholar
Abstract
Computer vision technology that automatically steers a passenger vehicle to keep it in its lane on a limited-access freeway is described. This system, LaneTrak, uses a forward looking monochrome CCD camera and a real-time image processing system on-board the vehicle. The image processing system identifies the lane boundaries based on lane markers and road edges. A pair of Kalman filters uses this information combined with vehicle dynamics sensor data to determine the vehicle angular heading error, lateral offset from the center of the road, and the lane curvature ahead. A preview proportional-integral control system determines the desired steering angle to keep the vehicle in the lane. The steering angle is then passed to another controller which activates an electric steering mechanism that positions the front wheels of the vehicle. We verified the full operation of LaneTrak with computer simulation before the system was designed and used a driving simulator to evaluate drivers' reaction to system failure. LaneTrak has been successfully demonstrated on a test track and on an unopened section of a limited-access freeway at highway speeds.

This publication has 0 references indexed in Scilit: