Zebra Motionworks™ for FRC 2022

It has been quite some time since we have made an update on this, March 12th 2020 if we are going by our support Slack. We have been working quietly to continue with this for the game this year. Due to the obvious chaos caused by the pandemic we were unable to finish our rollout in 2020 and and have had to scale back the 2022 rollout. Big thanks to all the teams and events that participated in the events for 2020 and a big shoutout the the crew that ran the PNW blue trailer system for the least work needed by us to refurbish the system.

We are glad to see R103.C and R707 in the game manual again this year. We are currently in discussions with 2 districts and 1 RPC to make the systems available for full coverage of their events. More information will be available and some announcements made when systems arrive at their locations and details finalized.

This years field is looking like a great candidate for system accuracy. Like the last game we will not be tracking the Z dimension so how high a robot climbed will be out of scope.

41 Likes

Very cool to see this back for 2022!

Any chance you can name the events/regions yet, or are you waiting for things to be finalized before doing so?

I (Matt the mentor) don’t want to jump the gun on any announcement those organizations may wish to make. Though I did want to let the community know we are still moving along with this. The pandemic didn’t kill the drive.

6 Likes

Do you guys provide the animations of the robots moving around the field, or is that someone else plotting your data?

I thought those animations were amazing and it was really informative to be able to have the students analyze where cycle time was being wasted by watching those animations and thinking about different strategies that could improve the cycle times. We also took an opportunity to plot some of the speed data to see how often we were hitting our max speed versus accelerating and used that to make an educated decision to change our drivetrain gear ratio to increase our top speed a bit.

Obviously our team took the raw data and put it to good use using some not so sophisticated tools (excel). But I was surprised how few teams seemed to avail themselves of this data. I would think that more teams would look at this data and analyze their own performance if there was an an easy interface to look at this data. I could also see this being a useful way to generate some scouting statistics as well. Any chance of integrating some data analysis tools into a interface that the teams could access through a web page for a particular event?

Really anyone is free to get the data and use it for whatever purpose they want. I am glad to hear that you and your team made good use of it.
All of the data gets pushed to the Blue Alliance who have graciously agreed to host it.

As a totally random example see this match from 2020 in PNW:

Just scroll to the bottom and you can get two visualizations, one that shows robot animation and the other that displays robot heat maps.

4 Likes

Once game pieces are tracked, this will revolutionize scouting similar to it’s impact on the NFL.

3 Likes

Wait, we still havent figured out if RFID is GOOD or BAD, how can we track them?

16 Likes

Yeah, those animations and heat maps are great. We actually pulled our data down and, after a little bit of manipulation were able to read it into Excel and then plot it versus time. We were then able to take the position data and, with a little bit of high school math, plot our velocity versus time and our acceleration versus time. If we had wanted to look at other things that could be derived from the data such as cornering speeds, or sudden stops (indicating impacts), we could have done that as well. The process of manipulating the data was time consuming enough that we really only looked at about half our matches from our single event in 2020. But from those half dozen matches we were able to see that there were times where our speed leveled off at a speed that matched pretty well with our drivetrain simulations. So, it was a nice validation of that. And, as I said we surmised that if we changed our gear ratio, we would be able to achieve higher top speeds resulting in an overall improvement in our cycle times.

I guess my request was to build a fairly simple interface that could a) pull the data from TBA, b) provide some pre-canned plotting functions like speed, acceleration, direction, etc. in addition to the animation and the heat maps. You could plot all the robots in a match on the same plot with the ability to turn robots on and off (just like you can with the animation). I’m sure as this data becomes more available, someone will develop a tool and share it with the rest of the community.

2 Likes

Does TBA api expose this data? (I haven’t checked lately)

1 Like

Yes.

ive been working on a version thats much more user friendly, hoping to have it ready before comps start

if you want to grab it yourself, the tba endpoint is /match/{match_key}/zebra_motionworks

You saw my whitepaper on that, right? Depends what you’re doing, and whether you have 20 or 0 years of experience.

3 Likes

Cool. Lemme know if you need anything. Otherwise ima go back to trying to write queries to identify what teams were actually doing based on their movement…it’s not going too well.

Main thing is just figuring out how to make a UI that doesnt look like it came from Windows 98. I think i found something? Will have to see if it is anything worthwhile though…

2 Likes

The obvious question is cost, but the NFL has already proven Zebra Motionworks works to track game pieces and players. Anyone sitting on their couch during the game can pull up stats like average time to throw, % of throws in tight windows, blitz %, completion percentage over expected (CPOE), rush yards over expected, air yards, % of time a WR is pressed, pressure rates and so on. All of that data comes from these same tags.

1 Like

It’s not entirely cost. Each NFL football is specifically designed to contain one of these active RFID tags so that it does not add to the overall weight of the ball nor does it interfere with the characteristics of the ball and how the ball acts. The tags we use for FRC come in at about 1.5 ounces. On this year’s game piece, if it was not perfectly centered inside that game piece, it would adversely affect the characteristics of that game piece. Also, although each NFL team brings 30+ NEW NFL regulation footballs to each game , only ONE is on the field at a time. There are 22 game pieces on the field this year, with who knows how many spares sitting around the outside. Other years, that number is even worse and the pieces are even smaller.

2 Likes

In short, Steamworks balls bad bad very bad, 2008 or 2014 balls maybe?

9 Likes

And even in those years, we still would not be able to accurately detect scoring.

We finished with refurbishment of the systems and should start the shipping process before the week ends. While going through the tags to replace lost / damaged ones I came across these from IRI 2019. They are the only two tags of the 1000+ we have that have a team number on them and I can’t bring myself to rip those markers off yet!

For those just curious the manual to setup this system is available here: Zebra’s MotionWorks Hardware and Software for FRC Setup Outline - Google Docs

20 Likes

Looks like all systems have been delivered to their locations (or delivery is ready to be scheduled by the receiving party). For fun, here are some shipping carnage pictures :frowning:

Yay for careful packing

2 Likes

SBPLI 1and 2???

4 Likes