Here are the latest updates for Limelight. The 2018.3 change log was only sent to our email list, so we want to include it in this post.
2018.4 adds new contour sorting options. These are fairly important for cube tracking this year, as teams don’t necessarily want to track the largest cube in view. In many cases, teams want to track the cube that is closest to their intakes. Many users have had to use the raw contours feature to implement their own sorting, so we want to make this as easy as possible.
Contour Sort Mode**
Select between “largest”, “smallest”, “highest”, “lowest”, “leftmost”, “rightmost”, and “closest” sort options.
We suggest trying the “closest” option for cube tracking
2018.3 addresses a major networktables reconnection bug which would cause NetworkTables settings changes to not propagate to Limelight. Thanks to Peter Johnson and the WPILib team for pinpointing and fixing the underlying NT bug. This was (as far as we know) the last high-priority bug facing Limelight.
Settings changes such as ledMode, pipeline, and camMode should always apply to Limelight. You should no longer need workarounds to change Limelight settings while debugging, after restarting robot code, and after rebooting the roborio.
Fix major NT syncing issue which broke settings changes (ledMode, pipeline, and camMode) during LabView debugging, and after a reset/reboot of the roborio.
The eye dropper wand uses the same 10 unit window for Hue, but now uses a 30 unit window for saturation and value. This means that thresholding is more often a one-click operation, rather than a multi-step process.
Setting the snapshot value to “1” will only take a single snapshot and reset the value to 0. Snapshotting is throttled to 2 snapshots per second.
Snapshot limit increased to 100 images.
Snapshot selector area is now scrollable to support 100 images.
We were having trouble pulling up the limelight feed from a IP in the web browser. What IP should I be using and is it possible to use this method at a competition? We were trying to 1uickly connect our limelight in the pit and were just gonna have a web browser with the feed during the match.
We too plan to use Chrome to stream the video feed Its found at http://Limelight.local:5800
I seems to use a bit more bandwidth over just the smart dashboard viewer. but we are only at 25% so we are good.
edit: If you can get to the web interface its printed there on the top Right above the live view.
I would like to see some of these cube tracker pipelines in practice. To bad our angle is 8" off the ground. Sounds like people who have been successful are using a looking down view. Our drivers sill like the bright yellow/green blob that shows up on the driverstation simply using the camera mode in drive.
I don’t know the details. The students moved some sliders, clicked some buttons, and now we can track cubes. They also worked pretty hard on the integration to get the drive train moving in the right direction, and have further iterated on the UX to make it usable during a match (particularly with figuring out which cube in a row to go for). The dashboard during a match is just Chrome to the camera stream URL with target boxes; it’s all on the robot from there.
We have a couple of ideas on how to solve the wrong-cube tracking, but they’re all UX-based atm. We may incorporate some form of sensor fusion for worlds. IMO these are the ‘fun’ problems that getting a Limelight enables us to experience & iterate on. It’s definitely not a silver bullet to high performance.
I know how Lime Light works I Beta tested it last fall. It is an impressive unit that keeps delivering. Its already out preforming its original design. I would love to see it in action with the cube solution.
We’ve noticed that the limelight web pages use the google fonts api to load Roboto, is there any way this can be migrated to a font local to the device (by downloading roboto and putting it on the limelight’s web server)?
We’re having some issues where the page won’t load because we have no internet while connected to the robot and it is trying to load Roboto from Google but can’t.
I think we have everything in place for local fonts. Only trouble is finding a solid repro procedure. What does your setup look like? OS, browser, cache clearing procedure, etc.
Also, we just released 2018.5. It is a small update, but it’s important if you want to use snapshots at an event.