Limelight 2024 Software Updates

Limelight 2024 Software Updates

Limelight Documentation Upgrade

  • The documentation has been rewritten to streamline the setup process

Limelight AprilTag Generator

  • https://tools.limelightvision.io/ now features the first-ever comprehensive, mm-accurate online AprilTag PDF generator.
  • Select your paper size, marker size, and tag IDs to generate a printable PDF.
  • This doesn’t work fully on Safari yet.

Limelight Map Builder

  • Limelight Tools
  • You can now build custom AprilTag maps with an intuitive UI.
  • The default family and tag size have been updated to match the 2024 field.

New Hardware Manager

  • The Finder Tool is now the Limelight Hardware Manager
  • It has been rewritten from scratch. It now reliably detects Limelights, provides more useful diagnostic information, and does not require restarts to work properly.
  • Get it now from the downloads page

Train your own Neural Networks

2024 AprilTag Map and Note Detector

  • The map and detector model have been added to the downloads page and the latest Limelight OS image.

Limelight OS 2024.0 (2/6/24)

ChArUco Calibration Fixes

  • Our ChArUco detector’s subpixel accuracy has been increased. A reprojection error of 1-2 pixels is now achievable with clipboard targets and 20 images.
  • Using the same camera and the same target, 2023.6 achieved an RPE of > 20 pixels, and 2024.0 achieved an RPE of 1.14 pixels.
  • Input fields no longer accept letters and special characters. This eliminates the potential for a crash.

Out-Of-The-Box Megatag Accuracy Improvement

  • Before this update, Limelight’s internal Megatag map generator referenced the UI’s tag size slider instead of the tag sizes supplied by the .fmap file.
  • Megatag now respects the tag sizes configured in fmap files and ignores the size slider.
  • If your size slider has not been set to 165.1 mm, you will notice an immediate improvement in localization accuracy

Performance Upgrades and Bugfixes

  • Higher FPS AprilTag pipelines
  • The performance of the Field-Space Visualizer has been significantly improved.

Bugfixes

  • Apriltags in 3D visualizers were sometimes drawn with incorrect or corrupted tag images. Tags are now always displayed correctly.
  • “v” / tv / “valid” will now only return “1” if there are valid detections. Previously, tv was always “1”

A few notes

  • You can score notes in the speaker by looking at the centered speaker apriltag, using crosshair calibration, and using “tx” as an input into a proportional control loop.

e.g.

swerveTurnSpeed = tx * .005
  • If you need to shoot from several locations, first use the apriltag 3d point offset feature, crosshair calibration, and tx before you attempt to use your robot’s pose to aim.

  • You can also perform automated pickup, climbing, and amp scoring with simple 2D pipelines. You do not need to implement real-time path generation to play this year’s game, and you can build a competitive robot with a simple approach to vision.

60 Likes

I see what you did there…

Super excited to play with the updated software this week, thanks Limelight team!

6 Likes

I mean, really? It took me like 20 seconds to find another one

Come on, this is the second time you’ve claimed “first ever” something and I’ve had to call you out on it. Don’t pull a Vex 2020 and let marketing garbage tear down an otherwise good reputation.

4 Likes

I don’t think you can print tags with this tool.

I just found that an online aruco generator added apriltags in early September, but my first print isn’t accurately sized.

There still doesn’t appear to be anything that serves as a replacement for

  • Installing Java and Apache Ant
  • Cloning the Apriltag Generation github Repo
  • Building and running a command line tool

besides our tool. It was built out of desperation :slight_smile:

5 Likes

In all fairness, that post was accurate before your edit

4 Likes

Personally, I’d either use the FIRST provided tags or throw some python together in a colab notebook. I’ll double check it later but I generally trust matplotlib to do sane things. Took me 2 minutes to generate april tags there that at first glance, appear accurate in size (look, my printer is perennially out of ink it’s a whole thing)

Just gave off “it’s the only tag generator you’ll ever need” vibes and rubbed me the wrong way.

4 Likes

I noticed that there wasn’t any mention of upgrading to NT4. Are there plans to upgrade to NT4 and possibly utilize structs or protobuf instead of sending over a giant json string? Being able to use these in LimelightLib instead of having to parse the json would be much faster. This is especially beneficial to teams using multiple Limelights, where just parsing alone can take up 5-7 ms every loop run.

8 Likes

Is there any ETA of incorporating Global Shutter Cameras through USB and processing AprilTags with them?

3 Likes

Does anyone have any code examples of this? The 3D april tag info on the docs didn’t have any that I could find.

@jdaming
The gif shown in the neural network section is using the built-in neural network pipeline. This model was built with the new Colab tutorial to make sure everything works as intended

If you’re referring to note-tracking, you can use TX and TY to servo on the note with your drivetrain. The 2D crosshair calibration and sorting features are usually necessary while tracking game pieces.

@brandonmcd

Check this link. You can use this point of interest offset and standard tx/ty tracking to always track the exact center of the goal if you need a wide range of shooting locations

5 Likes

@Brandon_Hjelstrom Thanks for getting this update out and supporting the community. We still love this product. Best hardware and software around for FRC.

Can you update us on when a new batch of LL3s will be available? Andymark shows mid-Feb. Does that sounds right?

Thanks again. Y’all are awesome.

5 Likes

I’m software-dumb, so forgive me - We have a LL2+, will we be able to use the 2024 updates that don’t involve the specific hardware of the LL3 (assuming no neural networks, no google coral, etc).

there is a 2024 image for your Limelight so the answer should be yes.

2 Likes

You can see the note here all models support learning based with the Google Coral.

I am not aware of any released features (it is just “faster”) that the LL3 has that the others can’t take advantage of.

4 Likes

Is it possible to run MegaTag and the Note Detector off a single limelight? Our team only has one LL2+ but I would still like to have both field positioning and the ability to automatically drive to and pick up a note. Thank you in advance!

1 Like

You can create two pipelines, one which tracks notes and a second that tracks april tags. Then you can switch between the two pipelines on the fly depending on what you’re trying to do.

8 Likes

NT4 is probably not happening soon, but updates that allow you to skip json parsing are coming.

4 Likes

Is there a particular reason for this?

Have you tried swapping Jackson to another library such as FastJSON? Jackson, while extremely popular and well supported, is not known for its speed*.

It’s likely worth running this benchmark program on a Rio. Swapping LimelightHelper’s parsing logic is quite trivial for anyone who is this far in the weeds already.

In any case, I’m very much looking forward to not having to parse JSON anymore just to count the number of targets we see.

(*Not that it’s insanely slow, just not the absolute fastest.)

4 Likes

We are having trouble installing the update. On balena etcher, after we select the zip file, it won’t let us select “compute module” - it says "missing drivers.