My team is deeply considering using the Kinect for it’s depth sensors. It also comes with a regular camera so we wouldn’t need to use another. The question for teams who have used one in the past. Did you have any problems with the lighting in the playing field and what did you do to counter this? All info/opinions about Kinects are welcome.
I took some videos using the Kinect’s patterned light and the calculated depth value per pixel. I saw some odd artifacts when pointing it at the lights, when pointing it towards the floor and having something shiny on the floor. Specifically, it was the plastic free-throw for Rebound Rumble. PolyCarb I assume, similar to this year’s field. Anyway, the polycarb had a crazy depth value. I don’t remember if it was a distant value or a close up value.
I also had some odd reflection issues off the diamond plate and the driver glass.
When I wasn’t on an official field, I simulated some of this using a halogen work light and some polycarb scraps on the floor.
Greg McKaskle
From past experience, they have no problems on the field, so long as you avoid pointing them at high IR sources (see overhead field lights for an example). You will probably have issues if you have them in a field which is exposed to sunlight. They do not work in sunlight.
You should note the response to Q108 in regard to the Kinect.
There are no rules that prohibit the use of Kinect-like sensors, however please note that R43 requires that “Servos be directly connected to the PWM ports on the roboRIO,” which applies to servos contained within COTS devices.
Last year, my team used the Kinect to choose when to score in autonomous. We did it because of the hot goals. It worked pretty well in the shop, but didn’t work at all during the competition. I don’t know if it had to do with the people in the background, the lighting, or the quality of the Kinect. We ended up just using our backup autonomous to score right away.
The depth camera works on the field.
My team just used the IR camera for goal detection and tracking, but at the event, there were windows behind the goals, washing out the camera.
My experience is a combo of the previous two posts. The IR camera will work under many situations, and it will not work well in others. At least at the regional I measured, IR or near-IR radiation sources were abundant. I mean, …, the lights were hot, and there were IR reflectors around that could bounce that light into my Kinect.
All in all, I thought it worked pretty well, but if you aren’t careful where you point it, your camera will get an eye-full of bright stuff.
Greg McKaskle