XIMEA cameras feature prominently at Robotex 2012

Darth Mäger, built by Indrek Tubalkain (Electronics), Siim Viilup (Mechanics), and Ronald Tammepõld (Software) of the Robotics Club of Tallinn University of Technology, took first place.
“I just wanted to say thank you to XIMEA for the CURRERA-R camera and your support, which made this win possible,” Tammepõld said.

The second place team used two XIMEA xiQ cameras – one looking ahead and one behind the robot – to see more of the field at one time and improve localization.
Images of the two goals were fed into Monte-Carlo Localization algorithm (particle filter) for navigation, while the a solenoid coil gun to ‘kick’ the ball into the goal.

“We use the RAW RGB images from the cameras at half the supported resolution (640x512), which we can process at about 45 frames per second (fps) from the two cameras. The image segmentation is made in YUYV color space,” explained Priit Kallas, of Tartu University.
“We’re doing the conversion from RAW to YUYV using. Quality and performance of the Ximea cameras helped us build a very competitive robot with good vision capable of chasing after small golf-balls from across the field and we’re very thankful for the support.”

The Neve3 robot, built by members of the Robotics Club of Estonian Information Technology College, took third place. According to team member Valdur Kaldvee, “Neve3 used a CURRERA-R intelligent PC camera running Linux (Ubuntu) with OpenCV vision library.

The use of smart camera with Intel Atom processor eliminated the need for any external PC.
The main task for CURRERA-R was object recognition.
Custom software in robot was able to detect balls, goals and field borders.
Thanks to XIMEA's fast image acquisition data path, Neve3 was able to image the field 60 times a second.
Depending on the types of objects to be detected, the recognition program was running at speeds up to 40 frames per second.

This kind of speed and the camera’s overall low latency gave Neve3 a very fast response time.
It is very important when driving towards a ball and at the same time making small corrections to hold the ball in the center of image.
The custom software was able to calculate the size and distance to different objects from image.
CURRERA-R smart camera was used also for other tasks than computer vision.

All of robots high level control was also implemented in CURRERA Smart camera which was connected through RS232 to microcontroller based on Atmel AVR.
This microcontroller was in charge of reading sensor values and controlling motors and solenoid.
High level program in CURRERA-R took the result of computer vision and sensor values from microcontroller and based on this information made decisions what actuator commands to send.