This is not really a solution. This is just don’t like it? Then don’t come. Which does what everyone in this thread is doing and ignores that it is still a team’s decision to use zebra. It is not us on CD. Some teams feel it is a disadvantage. I think that the data would be better served complete. Therefore, without removing the team’s choice, I would say release it later. What you are proposing is just to have teams not come to regionals with them, which I think the majority of the FIRST community wants to be every one.
From our experience at Chezy Champs in the fall, the trackers are discs very roughly 1.5" in diameter, maybe 0.75" thick at the thickest, and the weight was fairly negligible IMO. They could be attached with zip ties or double-sided tape, and while it was not the most convenient to attach them to our bot, I would be very surprised if there was a team that absolutely couldn’t find a spot. I find it understandable that teams would want to skip that process in favor of doing something they find more valuable, like maintenance or auto testing, but it really shouldn’t be much of a hassle for the vast majority of teams, and the scouting data can be very useful.
For further reference, here is our bot wearing the trackers - the tracker shown is the white and red object right next to the radio case.
If you feel it’s a disadvantage, complain to HQ that you think it’s a disadvantage and ask that it be banned from events.
I think you’ll find that you’re in the minority on this–I would say a fair number of teams either don’t care or like the idea. Personally, I’m in the “don’t care”. If tracking where your robot went was important, we’d have a student doing that, and you’d find robots in the way. I’ve been part of a team that did that before (or tried to… when two robots both tried to hit the same moving target in auto, they hit each other instead).
At which point did I expressly state I felt it was a disadvantage? I said some teams see it as a disadvantage, and therefor do not put it on their robot. Of the top 8 teams at the midwest regional, a regional which used zebra, 2338, 1732, 2451, and 1756 all did not attach tags, and from what I understand it was because they felt it put them at a disadvantage. As such, they did not attach a tag to their robots. That was all that happened. The data from these teams would have been important to have, and I feel like if there was no disadvantage one or more of these teams would have attached the tags.
Also, you can get all the data the tags give entirely by watching matches and keeping track of where robots go. I just feel like the organization and specificity of the data given by zebras would be far more helpful at analyzing the game than doing it by hand, and so more teams using it = better for all teams in the long run.
'member when we could hit each other in auton?
I’m really trying to come up with some time where seeing where exactly a robot is/was in most modern FRC games. Sure I guess in theory knowing if they prefer a certain path between two points but I think a human watching is def better to see that (turns out people are really good at finding patterns)
And I say this as someone who spent way longer than I care to admit building a tool to let me write queries on match data with “find me matches where team X was defended for more than 3 seconds” using Zebra.
Agreed. It’s cool data, and perhaps making like an overall field heatmap or something would be, at the very least, interesting to the curious, but to be honest I can’t really see much data that Zebra provides that a human (yes, specifically one single human) wouldn’t be able to get with their own eyes and a notebook, and perhaps a stopwatch.
As far as releasing data post event, if you want it, I guess, but I suspect like 99.8% of people have never pondered questions like “Yknow, at Silicon Valley Match 43, where was 973’s robot at T=86 seconds?” or the like.
I don’t think thats how you’d use it post event, it’d be more useful for heat maps and more big picture things. Which you can get by hand but its a pain.
Sorry, I shoulda been more clear - I don’t find a competitive advantage in having that data. It has a lot of value for things like webcasts if we could load it live and provide a fixed full field camera.
“Which one is yours again?” is asked too commonly by spectators and this would help with that. It would also make post match commentary far easier to annotate as well as make some slick visualizations WAY easier to make.
I don’t think it’s much use competitively but it is 100% useful for the culture change aspects of the program (anything we can do to make it seem more professional can help there)
Lemme repeat myself:
ill give you $50 if any team says theyre not going to events because ZEBRA exists there.
Id agree if i didnt find it so hard to get someone to track it by hand accurately.
I’ve read your points already, repeating them does not make them any more or less valid. I only stated that you would get more data points on zebra because more teams would opt in. Of course there’d be outliers, there will always be outliers.
No team will opt out of an event purely for zebra, I don’t believe I said they would. All you’d do is remove a teams choice to do it.
You don’t have to track the full position at every second to get good data from a human, only the general paths they take. For example, this year, if you saw they were taking a route near identical to one you were taking, maybe it is not the best choice to pick that team.
And im pretty clearly stating that teams wouldnt for the same reason they arent now.
This still requires someone taking note of that data. If my team size is small enough to make scouting difficult, why am i looking to make it a bigger problem when a solution already exists?
You do not need to even put a person on tracking their position, just watching a few of their matches should give you a general idea. However, your point makes sense. But until Zebra is implemented at every competition in FRC, teams will always say no to it, for one reason or another. I’m just trying to offer up a way in which more zebra data can be gotten, which should as a whole help teams out.
Well, the obvious answer is require it. The maybe less obvious answer is make teams realize that it does not provide other teams with any competitive advantage whatsoever.
Let’s be clear here, a heat map ends up just giving you “optimal” paths at a high level of play. I know for a fact some higher level teams generate these early in their design process to understand how they need to be able to move/acquire game pieces.
Sure it would give you the route that team preferred but I’d hate to play defense on someone on a suboptimal path only to have them reroute to a faster path…
We had Zebra data at our 2020 event. Most teams (but not all) installed it.
I would say that the data really provides very minimal information that the scouting team would not have been able to collect anyway.
So far, the biggest benefit of having the data is to be able to analyze our robot performance after the event. We found that our robot was reaching its maximum speed many times during typical matches. Our drivers also reported this, but having quantitative data was able to confirm what the drivers felt was really helpful in understanding the full scope of our performance. Based on this data, we are considering changing our drivertrain gear ratios to allow higher speed so that the robot is not limited by the max speed of the current gear ratio. This has the potential to improve our cycle time, so in the end, this data could make us more competitive in the future.
It has also been interesting to go back and study the animations of matches to understand match dynamics. But again, these are things that are also quite apparent just watching match videos. So, I don’t think the Zebra data really adds anything so much as just allowing you to quantify what is already visible.
Zebra data does not give us most of the “important” match data like scoring. So, as a data set, it reveals very little useful scouting data about any team. It is an interesting supplemental data set and I would love to see it used at all events by all teams. I could see us running some automated scripts on the data to extract things like maximum speed and average speed and total distance traveled. But I don’t think that it replaces the vast majority of scouting data and it really will not fundamentally change how we scout.
I’m not sure why some teams were worried about “giving away” Zebra data by installing it on their robot. It may have been that there was the fear of the unknown as this was a relatively new technology and no one was really that familiar with how the data would be used. But now that we have seen what the data is, and what it isn’t, I think a lot more teams are going to be willing to participate.
So this isn’t the first time this topic has come up. I wasn’t at Midwest, but I have talked to a number of folks involved. If I could do a bit of synthesis of what’s been under discussion so far:
Zebra tracking (ie, realtime, aggregate robot position data over time) lowers the barrier of entry for teams to understand locations and tendencies of robots on the field.
Refusing to use the markers at an event intentionally keeps that barrier higher around just your robot.
Intentionally keeping a high barrier of entry is not a behavior teams should be involved in.
My own thoughts
The main “lowering” occurs due to the real-time nature of the data. Post-event doesn’t matter much, a single person with time on their hands could gather it by just watching match replay videos.
However, IMO, his lowering factor isn’t really that much. As others have noted, to really make use of the data, you still have to know how to interpret it into a strategy. It just helps take some of the “grind” out of the data gathering and analysis.
That being said, I know many folks (including @MikLast) and I aren’t necessarily on the same page about the last part of the theory, and I’m ok with that.
I struggle because there wasn’t a rule made about how the system should or should not be used. I worry that pushing for behaviors beyond what the rulebook requires is a slippery slope. I think it could open questions about whether certain defense techniques, chokehold strategies, scorched-earth picking, or other closed scouting data should be acceptable. I’ve got plenty of thoughts on all of these, and am not saying all are “a-ok-acceptable”. But a lot of them are in a grey area.
I can’t find a great place to draw a solid line in all these cases. Rather, I think it’s worthwhile to judge the cases by going back to the basics:
- Is the strategy legal, per the rulebook?
- Does the execution of the strategy concretely hurt other teams, to the point where their students would be de-motivated from future participation in STEM?
In the specific case of Midwest 2020, the strategy was definitely legal. For #2, I’m not sure yet. Again, not being there, I can’t speak to it, and I also am not willing to assume the de-motivation occurred. I have personal biases that lead me to believe it was executed well, but I have no concrete proof of it.
Ultimately, I want to see the technology deployed more. I don’t like barriers. But I want to see a rule that codifies it, so teams don’t have to worry about handling the nuances of dealing with #2.
I haven’t dug into the details of Zebra as much as you, so please correct me if these things aren’t possible.
I think Zebra has the potential to add some cool metrics or help automate some aspects of scouting.
Some examples that I can think of include:
–2017 Average time spent at loading station per cycle or cycle times in general
–2018 Maybe information on where they typically get cubes from or vault cycle times
–2019 Probably some defensive metrics. Compare number of pieces scored with time spent with a defender
–2020 Average climb time
Yes, all of those things can be manually tracked, but I’d much rather have an automated system to collect that data so our scouts be more than just stopwatches.
The challenge is (I assume) it takes time to develop the technical solution to collect those metrics from Zebra. Since it’s currently only available at a handful of events and teams can opt out, it’s really hard to justify spending the time on developing that solution.
The tags themselves are tiny. According to Zebra’s own documentation, they are 2x4x4cm in size and weigh 20gm (0.7oz), so the idea that even a maximum weight robot would have a problem is not really a consideration. Any robot could afford to have these installed without any loss of function or risk of violating size and weight limits.
Note that while the system itself might be able to collect Z data, the CSVs that are released only have 2D points. So you can’t tell whether a team took 30 seconds to climb, or whether they climbed in 5 seconds and hung there for 25.