I am a programmer on rookie team 5822 WolfByte. We are looking to order a camera or two for our robot.
We are not planning to use GRIP or another type or vision processing because we feel we do not have enough time to figure it out. Because of vision obstruction on the field, we are looking to use a camera to help our drivers see where they are going. We currently are using a Logitech C270 camera and are looking for something with a wider range of vision. What would you suggest?
Virtually every out there uses a Microsoft life cam 3000. If you need to buy more, the specs of that are far better than what your bandwidth is. If you need a wide angle camera, I would just find a good lens to put over the camera.
Our Microsoft life cam 3000 worked fine during development but largely would not work at 2 regionals when we were on the field. We were eventually told by the FTA that there is a known incompatibility between it and the field management system (FMS). I can’t remember if it was java related but we do program in Java. We are using it for human vision (not robot targeting). We had to go out Thursday night of a regional and buy the only Network Camera in stock (a dlink) but had numerous issues again on the field and last I heard we got it going somewhat. It would not work in IE or Chrome. Again I believe FMS + Java were involved in the problem.
We used it this weekend at our event. We had a few times where the camera feed dropped, but it was a simple fix. We just had to delete the camera on the JAVA smart dashboard and then add the feed in again.
We’ve been using a Logitech C920 it has worked wonders for seeing where we are going and lining up low goal shots. Once we attached a working camera to the front of our ball spitter we were making probably 50% more shots on average.
We are using it as a way for our driver to line up with defenses we want to cross, as well as line up with the low goal. We have a separate Ethernet camera for vision tracking and such. We also have another one of the Microsoft cameras that connects to a raspberry pi for a periscope. They are great cameras that we have been using for the past couple years and we have never had a problem with them. I highly suggest these cameras for any team that wants to throw one on some point in the competition season (this year we ended up drilling through the metal band at the end and putting a few rivets in to secure it). Right now they are around $22 on Amazon so in my opinion they are a great “bang for your buck” camera.
Has your team tried using the Microsoft LifeCam HD-3000 for GRIP or other vision processing on something running Linux (including the RoboRIO)?
I suspect you might view it differently if you did.
FRC11 is using the Logitech C920. We had Microsoft LifeCam VX-3000 years ago, had Linux issues with them, and went to Logitech and PS3 Eye USB cameras along with Axis cameras here and there over the years.
We’re using three Logitech C920 cameras. One for vision processing with GRIP, two for driver station feeds, all three running through a Raspberry Pi 2.
We struggled a bit during build to find a way to change the exposure settings via script on Arch Linux, but one of our students managed to get an old Logitech utility built and working on the Pi. Without that the image processing camera would have been very difficult to tune.
We also had issues at our first regional getting the two camera feeds to reliably show up on SmartDashboard, using static IPs. In the end we mistakenly had the “mDNS” option still enabled in the SmartDashboard, and the Pi had been forgetting its static IP when cold-booted. After fixing those the feeds were 100% for us.