Stereo Vision Sensor

The stereo vision sensor is the joint work of the working group "Dependable Embedded Systems (DES)" at Seibersdorf research and the Advanced Computer Vision GmbH (ACV), and is part of the Embedded Vision Applications initiative at Seibersdorf research.

The main task of the stereo vision sensor is obstacle detection in front of the autonomous vehicle in the range of 5 - 20m. The purpose of the obstacle detection algorithm is to identify a distance to a detected obstacle, for each direction in the field of view parallel to the surface of the road. The resolution in horizontal direction is 0.5°, the viewing field extends horizontally 20° to each side of the forward direction of the vehicle. The cycle time of the detection algorithm is 100ms.

Finally, the stereo sensor supports the extraction of the left and the right lane border of the road.

Features

The Stereo Vision Sensor consists of a pair of Basler A601f monochrome cameras with a resolution of 656 (H) x 491 (V) and a quantization of 8 bits/pixel. The focal length of the lenses is 8.5mm and the baseline of the stereo head is 0.35m. The cameras are connected by two 400Mbit- Firewire-cables to an embedded vision system, which is placed inside the vehicle. The stereo vision sensor head itself is mounted in front of the windshield of an autonomous vehicle. The embedded system is referred to as VisionBox and is based on a Texas Instruments TMS320C6414 DSP is running at 1GHz. It offers two 400MBit-Firewire- and one 100MBit-Ethernet-interface and is responsible for the synchronous acquisition of both images, for the execution of the computer vision algorithm as well as for communication with the vehicle via Ethernet. The operating system is DSP/BIOS from Texas Instruments.

The Smart Systems research team developed the embedded Stereo Vison Sensor - the intelligent eyes – for the vehicle of the SciAutonics team which took part at the DARPA Grand Challenge. An improved version of the sensor will be used by the team SciAutonics / Auburn Engineering for the DARPA Urban Challenge 2007 as smart eyes for their vehicle.

DARPA Grand Challenge Movie