I’ve seen in competition where the robot has like a camera with green lights around it to detect the goal like in Stronghold. Could someone direct me to what camera is being used or any alternatives. Thanks!
If on a budget, the Microsoft web cam from the KoPand a green LED ringwill get you there for about a tenth the price, and you can get good support through screenstepslive, CD and other FIRST fora. There are cheaper web cams, but you may not be able to get as much help.
You can also “roll your own” using an LED ring like this one from AndyMark, or these from superbrightleds.com (also found in FIRST Choice).
That being said, if you can afford and manage to get your hands on a Limelight, from what I’ve heard it makes computer vision a much easier problem to tackle.
EDIT: sniped
The LimeLight is a great product, but it’s certainly not cheap, nor for everyone. (Personally, if 2019 is heavy on vision like 2016 or 2017, we’ll be considering getting a LimeLight. But not every team will be able to fully utilize it.)
If you have a programming team who enjoys tinkering with things on their own, there’s nothing wrong with running a USB camera to a Raspberry Pi and making a mount for it with some space for LEDs. GRIP is really easy to work with.
1 Like
we have no affiliation to it?
973 was lucky enough to test a Beta unit, but we like it for it’s functionality… not because we were able to Beta test it.
I would definitely go with the Limelight if your team is new to vision. Getting the quality, reliability, and latency to the level of the Limelight will take significant time and effort. That time in my opinion could be better spent on developing code to make your robot actually use the vision information. It’s expensive but worth it if you plan to use vision during an FRC season.
Ahhh shoot. I got the Greyt 973 and LimeLight 987 (I think? maybe?) stuff confused. My apologies.
The LimeLight is great. But it ain’t cheap.
The LimeLight is great, but it aint GreyT.
Does anyone have an idea when the Limelight cameras will be available again?
I’m surprised to have not seen the Jevois mentioned here. It integrates the camera and processing together and is preloaded with the computer vision library openCV. It is capable of running ML algorithms as well. Since the processing is done onboard the camera, some latency can be eliminated. It provides the benefits of a limelight for much cheaper, around 50 dollars.
1 Like
Jevois is significantly more complicated to use than a limelight as well, and that is where the price difference comes from.
1 Like
Can limelight be used with java coding?
Sent from my Galaxy S8 using Tapatalk
I too have no affiliation to Lime Light and also were part of the Beta testers.
I suggest the Lime Light as well. The investment is steep but not wasting a lot of time with marginal success is priceless. The LL team has done a fantastic job developing a simple and fast configuration tools to many common vision problems found in FRC. If something new pops up they are quick to put out an update.
My team has used a mini PC, camera, LED rings, running GRIP, it worked but there were multiple points of failure and hours of configuration for one vision pipeline. Lime Light has minimum one POE-Ethernet connection and DONE, 9 selectable vision pipelines, and a USB port for optional driver camera. Also coming this fall I have heard it will support GRIP pipelines so if the built in vision tools can’t do what you want you can export one from GRIP.
Enough Fan boying here. I suggestion to “new to vision programmers” is
While you wait for your Lime Light to get approved
1 get a USB camera
2 install GRIP on a PC
3 run the tutorials found on screen steps for GRIP
4 now make a decision to try and
a create the pipeline to OpenCV code to run on the Roborio in your robot code.
b find a co-processor to put on the robot to run either the above OpenCV or GRIP Pipeline
c burn the network bandwidth to run the GRIP vision pipeline on your driverstation
d get your lime light or exsplore the multiple other vision solution like using an old Android phone
Now is the time to waist trying to get a lesser solution working. If you do it during build season its too late unless it’s Lime Light*.
*sorry more fan boy
The Lime Light post values to NetworkTables that your Robot Code can use to acquire a target.
1 Like
Last month, when I asked, they thought it would be October when they would have more.
With resources such as Grip and FRC-PDR, anyone with basic programming knowledge, a USB camera and an LED ring can make industry standard machine vision software.
As for camera suggestions: I like the genius 120 camera and any run of the mill IR USB camera for night surveillance (typically requires some easy hardware modification to make the IR LEDs always be on). IR cameras are especially cool because you don’t have to worry about RGB/HSV threshold tuning, and they allow for insanely tight threshold values on retro reflective tap (250-255)
Loads of issues with IR cameras from outside light sources and reflective surfaces, even with contour filtering. If you are playing in a venue that has sunlight, it’s a real challenge. This can work but this isn’t in the same realm of difficulty as using the Limelight or the JeVois.
This past year, my team used Jevois with an infrared LED ring for tracking the retro reflective tape on the switch. We never had any issues with sunlight interfering with the tracking. Granted, this may prove to be a problem if the field is completely flooded with direct sunlight, then the camera might get fooled by reflections. I definitely believe that in the correct conditions, sunlight can interfere with infrared tracking, but our team competed at a venue where the field had some sunlight on it, and we never experienced such an issue. Additionally, even tracking a color of visible light (such as the ever common green light) would most likely experience the same issue with reflections from sunlight. So I don’t feel that this issue is unique to infrared tracking.
Also, in order for Jevois to see infrared light, we had to do some surgery and slice the infrared filter off of the lens. Thankfully, Jevois now offers lenses without infrared filters. 
This past year, my team used Jevois with an infrared LED ring for tracking the retro reflective tape on the switch. We never had any issues with sunlight interfering with the tracking. Granted, this may prove to be a problem if the field is completely flooded with direct sunlight, then the camera might get fooled by reflections. I definitely believe that in the correct conditions, sunlight can interfere with infrared tracking, but our team competed at a venue where the field had some sunlight on it, and we never experienced such an issue. Additionally, even tracking a color of visible light (such as the ever common green light) would most likely experience the same issue with reflections from sunlight. So I don’t feel that this issue is unique to infrared tracking. But I haven’t done extensive testing
Also, in order for Jevois to see infrared light, we had to do some surgery and slice the infrared filter off of the lens. Thankfully, Jevois now offers lenses without infrared filters.