News
Vision Components is a leading international manufacturer of embedded vision systems. The freely programmable cameras with powerful onboard CPUs perform image processing tasks on their own without additional computer. Vision Compo-nents offers OEMs versatile Linux-based embedded systems for 2D and 3D image processing, supplied as board cameras or in protective casings. These are complemented by a growing range of ultracompact MIPI camera boards, which connect to a variety of different CPU boards. Software libraries and customized solutions are also part of the program. The team of experts can draw on extensive knowledge and over 20 years of experience with imaging applications. The company based in Ettlingen, Germany, was founded in 1996 by Michael Engel, the inventor of the first industrial-grade intelligent camera. Today, Vision Components works with local partners in over 25 countries to provide consistent customer focus and readily available expertise throughout the world.
Content Filed Under:
Industry:
N/A
Application:
N/A
Robot Vehicle Uses VC Smart Camera as its Brain and Eye
POSTED 04/09/2004
Summary: A team of students lead by Michael Armani relies on Vision-Components VC2028 ‘‘smart camera’‘ to be the brain and eyes of a unique robotic vehicle.
The U.S. congress mandated that its Defense Advanced Projects Research Agency, DARPA, would host an autonomous vehicle race to foster scientific development in the field of autonomous vehicles. The U.S. military wishes to automate 1/3rd of ground forces by 2015; such vehicles are meant to carry supplies to troops or perform reconnaissance tasks similar to UAV’s. The vehicles must be able to travel through 200 miles of rough desert terrain and weather. Team Armani was created to meet such a challenge.
The ‘‘DARPA Grand Challenge’‘ was held on March 13th 2004. 15 vehicles were allowed to compete. Within the first few minutes and only tens of meters, 8 vehicles broke down. Team Red’s vehicle made it the furthest at 7.3 miles, but they were using a backup Ladar – (laser distance find system) which could not provide the resolution needed to spot a small rock which broke their vehicle. The teams that made it the furthest, such as team Red, SciAutonics II and DAD, all were using camera systems to augment their other terrain sensing systems.
‘‘The Grand Challenge demonstrated the need for truly intelligent camera systems. That is where the VC2028 comes in. Our teams’ approach is to use the Vision-Components smart camera to see and process the speed at which objects move past the vehicle, thereby creating a 3D obstacle map. Then the VC2028’s DSP chip will also be responsible for processing the 3D data to find the best path.’‘
There are few choices for sensing and processing data from the environment, and the VC2028 as an integrated solution seems to be the most appealing. Primary alternatives include connecting several delicate and expensive CCD cameras, RADAR systems, or Ladar systems to bulky hi-end computers; all of these combinations may cause a reliability issue because of the number of components needing power, cooling, vibration isolation etc. The VC2028 is a powerful solution because it does not require cooling even in the desert, its’ sturdy packaging prevents moisture and shock damage, and most importantly it carries a Texas Instruments DSP chip allowing for the equivalent of 1GHz of computing power. The VC2028 is also an industrial camera, so it is compatible with standard lenses and stands, and the DSP programs can be copied to other VC cameras in only seconds.
Team Armani is making significant progress towards completing their autonomous vehicle. Pictured above in front of other robotic vehicles is our prototype, and the VC2028 is mounted at the top of the vehicle. So far the vehicle has been tested with autonomous path following and power balancing algorithms, while the most daunting task that remains is the obstacle detection. In the first month of work, Team Armani was able to configure and program the VC2028 computer interface, overlay display, user defined image processing routines while taking images, real-time particle tracking, shutter control and focusing algorithms onto the DSP. The final challenges remaining are to optimize to programs for real-time speed and to test the desired algorithms, while another team works to turn an ATV into a robotic platform. Team Armani hopes to enter the newly founded International Robotic Racing Federation racing in October 2004, and plans are in store for another DARPA Grand Challenge.