Formula Electric Belgium is preparing for driverless competition with dual head stereo vision camera from GeT-Cameras.
The Formula Electric Belgium team, consisting of 25 Belgian university students, are building their first autonomous Formula car. In the previous years, the team joined in the electric competition along with more than 50 other competitors. This year they decided to enroll in the autonomous competition once more to inspire people with the possibilities of electric and driverless cars.
Their Formula racecar competes in multiple disciplines, acceleration, grip strength and endurance. With exactly 1000 collectable points, the team strives to win this year’s competition.
For acceleration, the car has to get from standstill to the 75-meter finish line as quickly as possible; Next, the car gets tested on grip strength by driving as fast as possible on a short round track. Lastly the endurance test, by driving a lap of 22km. Besides the competition, the students get graded on their business plan and their cost analysis, which is where the fairly priced cameras of GeT-Cameras came into play.
Due to the well documented and easy to use software, the team swiftly implemented the dual head stereo vision camera from GeT-Cameras. The software of the dual head camera automatically recognized the two image sensors. As a result, the team quickly gathered their first stereo images. With these stereo images they are able to generate a 3D image.
Since the formula car is driverless, has to withstand high speeds, G-forces and extreme weather conditions, the camera must see everything. Therefore, the team decided to go for a superior camera with a global shutter. A global shutter camera captures crisps images without blur, therefore suitable in these harsh conditions.
The casing for the camera was 3D-printed with a hydrophobic glass in front of the lens to make sure that raindrops won’t disturb their readings.
The performant image sensors of the camera captures enough light to achieve short exposure times, this is needed to achieve sharp images at high speed driving. To push the boundaries further, the team uses digital gain without introducing too much noise; Therefore they manage to use a short exposure time up to 2 milliseconds.
Open source software
As appreciation for our collaboration and sponsoring, the team will release their 3D software open source. This means you can download their software and immediately get an idea of how to use the cameras in a 3D-setup.
To calculate a 3D image using the stereo vision camera, the following steps are required:
Step one: Calibrate the cameras, make sure you have 2 cameras pointing at the same calibration sheet, with the same capturing conditions.
Step two: calculate the cameras position, the software will figure out where the cameras are in comparison to each other.
Step three: determine what object you want to recognize.
When the cameras are calibrated you are able to generate a 3D image. The next step is to determine what you want to see. As soon as you can identify and locate an object in both images, the software can triangulate it to provide a 3D location. The detection of an object has to be implemented by the user; Although a simple example will be provided.
Unfortunately, due to the corona virus, the students got some major setbacks. Firstly, to ensures their health, the university prohibited them from working in pairs or at the same location. Therefore, the production process of the car got delayed. Secondly, the competitions got cancelled. Despite these setbacks, the team is working hard to minimize the impact of the Lockdown. Whatever they can make at home, they do, like the printed circuit boards, software for the different computers on board and more extensive simulations.
The 3D vision software is available! Read more about it on GitHub.
You can read more about the Formula Electric Belgium on their website.
Make sure to follow GeT-Cameras on LinkedIn, so that we can keep you up to date about our new products and blogs.