It is with great excitement that we announce the successful Beta of DataZ for FRC at this past weekend’s Half Hollow Hills Invitational. Congratulations to the winners and finalists at this exciting event! We extend our thanks to FRC Team 3624 for allowing us to run this Beta and additional thanks to them, as well as FRC Team 329, for their assistance with system setup and logistics. We of course cannot forget to thank all of the teams who participated. We had 100% participation at this event! We also would like to thank the FTA for allowing the Beta to run. We operated under the condition that if at any time there was any question of field interference, DataZ collection would be halted. We are pleased to tell you that there were absolutely no issues. As far as Betas go, this was exceptional!
Here are some of the interesting takeaways from this Beta:
- Tag handouts could be smoothed out. It took some time to hand out our prepackaged and activated tags to teams. Much of this time was introducing the system.
- Tag collection methods. We are pleased to say that of the 54 tags handed out during the day, all 54 found their way back to us at the end of the event.
- System setup. We are happy to inform everyone that system setup is smooth, quick, and can be done at the same time as field setup. Our current infrastructure requirements are non-intrusive and we received no complaints about field obstruction or match flow / queuing.
- Empty field testing vs Full field testing. No dead spots were seen on the field, though mild distortion when tags were less than 12 inches from the diamond plate driver station walls. This can be improved through receiver location, or adding additional receivers.
- 27 robots, through all Qual matches, 1 replay, and all Playoff matches each with two tags, generating over half a million data points!
- 2 robots were not tracked for a single match across the whole event, one didn’t have time to install the tags before their first match, the second had an issue with tag placement. After those two issues, all robots remained tracked.
- One tag fell off throughout the whole event. This tag was mounted to a 3D printed part which snapped off the robot. The tag was still attached to the plastic part. This helped us guide teams to proper tag placement locations.
- Our current data processing methods could be improved. Minor bug fixes in current FRC oriented data viewer and system setup applications.
Important next steps:
- Sleep! It was a long and exciting weekend (as you know FIRST events are) We took Sunday off.
- DataZ for FRC results (Heatmaps, Videos, and Stats) will be released throughout the week!
- We have been in communication with a local regional about the use of this system for the 2018 season. They appear very interested.
- We are expanding the number of teams we are working with to help discover the full potential of the data that this system provides.
All videos will be posted along side our teasers on our YouTube channel: DataZ 4FRC
Additional information can be found on our previous threads:
DataZ for FRC
and
DataZ for FRC: Bench Grinder
This is really awesome! Are there any plans on releasing all the raw data for teams to play around with? Also do you envision this data being accessible real time for teams to use on the field or only after each match?
Thanks! We are looking into ways of distributing the raw data in the future. As for real time accessibility during a match (on field) it would probably require integration with FMS to some extent. The system can provide location information within ± 4in and using our current processing has about a half second delay though that could be improved. This might be useful but probably would not be a replacement for other autonomous control styles. The collection is quick enough to be compiled and evaluated inter match and so could be available for scouting applications as well as “live” video stream overlays (if the stream is on a >.5s delay). The NFL uses this type data to discuss plays and statistics during the time between downs or timeouts. We think this could be an interesting addition to the current activities between FRC matches or field timeouts.
This is super exciting! I’m already working on ways to make use of this data, so I would also love to see a download for the data gathered at Half Hills. How will teams be able to access this at events?
Hi Sam, We are glad to hear you are already kicking around ideas of how to make use of the data. We encourage anyone with ideas to please post on the thread or PM us as we would love to have community input. We are still working though the data ourselves (given the holiday week) and some processed graphs and stats will be posted tonight along with how to access them and samples posted to this thread or on a new one. We learned a lot about the differences between 6 robots on an open field in a large area and 27 robots in a gym surrounding a full 2017 field. We are working to make this raw data available to everyone. As for availability of DataZ data at events we are weighing our options. We have identified a few internal (Zebra) resources we might be able to utilize as well as some FRC community base options that we could reach out to. Similarly, if people have suggestions on where they would like to see DataZ (Raw Data or Processed Results) let us know and we can look into it!
Congratulations on the successful pilot demonstration applying this exciting technology to the FRC competition. We are glad to have participated. We found application to be painless and unintrusive. The potential to competitivess and entertainment is great- the shown tracing and heat maps are just a taste of what can come.
As we discussed at HHH (and now augmented), I’d recommend the capture and sharing of the presently defined data for each individual robot per match to individual .csv files, posted as rapidly as possible for teams to download for analysis. The system should be augmented to allow user defined telemetry from the robot to be stored per record to the same file. Perhaps pings times to the robot could included per record as well to aid in time correlation between user data and DataZ data. Per our discussion on whether DataZ should re-sample and interpolate all data to be aligned to regular intervals, perhaps both could be made available, as separate .csv files, or be interleaved yet applying tags per record for easy separation (e.g. grep). Perhaps .csv files with only DataZ data for all robots could be public, and then user data augmented .csv files could be downloadable only by the registered owning team.
I also like the idea of live augmented video feeds as you’ve described, but also like DataZ and possibly user data telemetry as well, such that driver teams might customize displays to themselves to use during matches :eek: They’d only be limited by their creativity and bandwidth allocation.