Vision for Longitudinal Vehicle Control

Philip F. McLauchlan and Jitendra Malik,
EECS Computer Science Division, University of California at Berkeley,
Berkeley, CA 94720, USA. Email pm,malik@cs.berkeley.edu

Abstract:

An important component of the drive towards intelligent vehicles is the ability to maintain a fixed distance from a lead vehicle using feedback provided by range sensors. We are investigating the possibility of using stereo vision to provide the range information, in conjunction with a scanning laser radar sensor.

The vision algorithms build on fixation and reconstruction algorithms designed for active vision systems, and combine stereo and motion cues. We shall present preliminary results comparing the quality of range measurements provided by a vision system with the laser radar system, using data obtained off-line. Later we will implement the tracker in real time on a network of C40 DSPs, and combine the laser and vision sensing in a cooperative manner.

• 1 Introduction
• 2 Visual Tracking
• 3 Fixation and Scene Reconstruction
• 2D affine reconstruction
• 5 Affine transfer
• 6 Robust stereo and temporal matching
• 7 Results
• 8 Discussion and Conclusions
• References