Real-time stereo vision system guides land vehicle

At the department of computer engineering of Parma University (Parma, Italy), Alberto Broggi and his colleagues have successfully demonstrated a stereo-vision-based system to increase road safety on an autonomously controlled land vehicle. Called Generic Obstacle and Lane Detection (GOLD), the vision system is based on a Pentium processor capable of processing 25 frames/s.

Aug 1st, 1998

Real-time stereo vision system guides land vehicle

At the department of computer engineering of Parma University (Parma, Italy), Alberto Broggi and his colleagues have successfully demonstrated a stereo-vision-based system to increase road safety on an autonomously controlled land vehicle. Called Generic Obstacle and Lane Detection (GOLD), the vision system is based on a Pentium processor capable of processing 25 frames/s.

"The system has been tested both in the laboratory and on-board the university`s experimental land vehicle. It has demonstrated robustness with respect to shadows and changing illumination conditions, different road textures, and vehicle movements," says Broggi.

To localize obstacles in front of the vehicle, pairs of stereo images are processed using inverse perspective mapping. This technique is based on a transform that remaps both stereo images into a common domain. Any disparity in the remapped images is then due to a deviation from the road model, thereby allowing the detection of potential obstacles.

Lane detection is based on a pattern-matching technique that relies on the presence of road markings. "In the remapped domain, the detection of lane markings is simplified because markings can be represented as almost vertical lines with constant width," says Broggi. And, because the remapped image represents a bird`s-eye view of the road surface, lane-marking detection can be accomplished through morphological processing.

"Because both obstacle location and detection of lane markings are based on the processing of images remapped into the same domain, the fusion of the result of the two operations is straightforward," Broggi says. When one or more obstacles are detected, their position and size are given as inputs to the lane-detection algorithm. The obstacle area is not considered during the lane-detection process, thus avoiding the risk that the obstacle shape could be confused by a part of a lane marking.

Output of the image-processing stage is displayed on both an on-board monitor and a control panel to give visual feedback to the operator. The university`s experimental land vehicle has been driven for more than 3000 km along urban roads and freeways at speeds to 80 km/h. The maximum speed tested was 123 km/h.

More in Consumer Packaged Goods