NZ In The Zone Scrimmage Thread

VEX Robotics Competition Discussion VRC > In The Zone (17/18)
1

Welcome back to another season of NZ scrimmages.

This season we’re going to try something a bit different and share a lot more statistics and probably less actual details. We’ll be looking at ways to increase the amount of data we can gather- beyond what TM offers. For the most part @Jij is going to be writing up the summaries, he may even share some of the warnings he dished out as head ref. Neither of us will be sharing any pictures or video, so please don’t ask. If there’s a metric that you think would be interesting, let us know and we’ll see if we can add it in for later updates.

This season is slightly irregular compared to last year. The first scrimmage was a workshop, but was referred to as “Scrimmage 1”. This means all are actual scrimmages are a number out. (Eg. Our first actual scrimmage is “scrimmage 2”). Also, Kiwibots will be hosting APAC at the end of the year, which we should be able to gather a huge amount of data from.

Current scrimmage dates are as follows. Please allow a couple days after the event to analyse the data.

Scrimmage 1 – June 10th 2017 (Workshop)
Scrimmage 2 – July 1st 2017
Scrimmage 3 – July 29th 2017
Scrimmage 4 – August 12th 2017
Scrimmage 5 – August 26th 2017
Scrimmage 6 – September 9th 2017
Scrimmage 7 – September 23rd 2017
Scrimmage 8 – October 14th 2017
Scrimmage 10 – February 10th 2018
Scrimmage 11 – February 17th 2018

*I am aware scrimmage 9 is missing, this is all the information I have currently.

At this time, I’d like to thank Hayden for making these events actually happen and in general organising the Auckland region. Also, a big thank you to AURA, who will be, pretty much, the only source of volunteers we will see in Auckland. Without the combined work of both, none of this would be happening.

So, sit back, relax, and enjoy your irregularly scheduled statistics.

2

Just like last year, we’re going to give out a bunch of statistics and analysis of what’s going on at our scrimmages here in Auckland. Whilst we won’t reveal the robots (that’s not our secret to share), we’re quite happy to describe them by numbers. Last year we got a few things wrong but hopefully we produced some interesting data. This year we’re turning it up a notch, fixing a few of our silly errors such that it makes a bit more sense, and making it a whole load more graphical (instead of just copy/pasting some results straight from a script).

With only 21 teams competing in 37 qualification matches and 11 elimination matches, we don’t have a great sample size to play with. It’s also true that this little sample isn’t representative at all of Auckland, with robots that didn’t work as planned and many teams not ready to compete yet. Comparing scrimmages with each other won’t necessarily be great, especially with this “odd” scrimmage. We may adjust the statistics displayed as we find more interesting and relevant bits and pieces.

We’ll kick off with looking at the match scores overall:

The median qualification match score was 16 pts for a single alliance, or 35 for the combined total score between alliances. As expected, the scores for the elimination matches tended to be higher, with a median of 42 pts for a single alliance or 84 pts for the combined total score. If we’re being honest, these aren’t great scores. Having said that, this was also only the first scrimmage, with a lot of ambitious robots that weren’t quite functioning as intended, so expect a lot of improvement.

Looking deeper into the scores, we decided what’s probably more interesting is where the points are coming from:

As immediately obvious, the scales of scores increased across the board between the qualification and elimination matches. Cones still contributed roughly a quarter of the score, with mobile goal scoring increasing from about 50% to 60%. Highest stack bonuses still contributed about 10-15% of the score, with parking at less than 10%. There was only ever about 1 or 2 highest stack bonuses awarded, indicating that most scoring occurred only on a couple of goals.

Most mobile goal points came from the 10-point zone. Obviously, as per the game manual, only 1 mobile goal can be scored in each alliances 20 point zone – an important fact to keep in mind. Placement of the mobile goals is better showed by the following graph:

Looking slightly deeper into the cone scoring, we can get a distribution of the number of cones scored in each match:

As you’ve probably already figured, not many cones were scored. The median number of cones scored by a single alliance in a qualification match was 1, or 4 between both alliances. For elimination matches, we saw this increase to a median of 5 for a single alliance, or 8 between both alliances. A significant increase, but still low numbers for all.

To finish off, we’ll look at the improvement over the course of the day, using a 5-point moving median of the combined match scores (scores fluctuated significantly):

This is a pretty rough graph. However, it does indicate a noticeable improvement as robots were fixed and adjusted throughout the day to be better and better. This isn’t normal - ideally, robots should turn up to an event ready to go without modification, so this is something to bear in mind when reviewing the data from this event. Here in New Zealand, it’s rare for teams to own their own field. Instead, until now, the vast majority of teams have only been working with the one cone each they received about a month ago.

Other statistics:

  • No one won the autonomous bonus during the qualification matches
  • No attempts at the skills challenges
  • Highest single alliance match score was 78 points by 2900D and 2900A
  • Average number of robots parked per game was 0.9
  • Median difference in qualification match scores was 7
  • Guesstimated highest stack was about 7 cones tall
3

Just add a few extra details about the teams competing and the scrimmage itself, that don’t fit into the stats side of things.

  • There were no teams that won all their qualification matches.
  • The first, third and fifth seeded teams were all from Onehunga High School
  • The second and fourth seeded teams were both from Glenfield College
  • For eliminations both Onehunga and Glenfield allianced with themselves
  • Finals were played across 3 matches. Onehunga winning the first, Glenfield playing heavy defence the second and taking it marginally. Onehunga coming back for the third.

Our first few scrimmages are usually very relaxed in terms of rules and match times. We’d much rather get teams used to the game before enforcing rules. This is why you won’t find them on RobotEvents.

For those running events in the future, resetting isn’t as bad as it looks. There are a lot of objects, but they’re fast to reset. It’s the same with scoring. It takes a bit of time to get used to what you’re looking at on field, but once you get the hang of it, it can be done fairly quickly. I mentioned in the initial post about trying to gather more data, how we score the match is part of this.

I’d like to thank Glenfield College for hosting us and helping out with some of our missing equipment. To the guys from AURA that showed up to help (in the middle of exam season), I’ll say it again, these events can’t happen without you. So, thank you.

4

Very interesting. Thanks for the statistics. It would be so informative to see these statistics as your season progresses.

I have a question - what was most of the time spent doing? I notice that most of the scoring was mobile goals with few cones. So, can I infer that means that the majority of the time was spent mocing mobile goals? Or were attempts made to score cones but they just weren’t that successful? Also, what about the driver loads? Did anyone bother with them at all?

5

I really like the idea of putting out all of these statistics. It’s nice to get a better idea of what to expect in terms of scoring. Looking forward to the upcoming scrimmages. Thanks for taking the time to put them together!

6

Wow, @Jij, thanks so much for taking the time to write up these statistics! This is a lot of useful info!

7

I think the best answer to this question is say a lot a teams were under prepared. I think that’s mostly a byproduct of our relaxed approach at early events. What we saw on the day was teams that had planned for more complex designs showing up with more simplified (see: non-complete) versions. Add that to a lack of practice, as the moving median shows, and you get a lot of slow or non successful attempts at scoring cones. When it’s easier to push goals around and it’s scoring more, at this point in our season, it became the chosen strategy. I will add that a lot of these robot were not designed to be able to push goals, so it took some time.

8

29 July 2017 Auckland Scrimmage Summary

In what was our second VEX In The Zone scrimmage here in Auckland, 28 robots turned up to compete and test out their robots at Kristin School. With 49 Qualification Matches, each team got 7 games to play before heading into the Alliance Selection.

The first ranked team at the end of the Qualification Matches was 7682E (Wingus & Dingus), who were the only team to win all their games and one of only two to win 6/7 of their autonomous bonuses. In second place was 2931D closely followed by 2931F (both from Pinehurst School). 2931D had 6 wins and a tie with no autonomous points, and 2931F had 6 wins and a loss, with 20 autonomous points. If the second ranked team (2931D) had lost that game which had been a tie, they would have ended up in 6th position – in this event, every point was crucial.

For the Elimination Games, we saw the blue alliance (lower seeded alliance) beat the red alliance in every playoff except for the case with the 1st seeded alliance, who progressed smoothly into the Finals. Hence, we had a Finals between the 1st seeded alliance (7682E, 2900A) and the 7th seeded alliance (2900G, 2915D). The first seeded alliance won the finals with two convincing wins. Congratulations 7682E (Wingus & Dingus) and 2900A (SymbiOHSis/Onehunga High School)!

Time for some statistics.

At first glance, these match scores look fairly similar to what we saw at the scrimmage on 1st July. The mode of the single alliance scores was still less than or equal to 10 for the Qualification Matches. Overall, scores bumped up a little, but not significantly. The median Qualification Match score was 19 pts for a single alliance, or 42 for the combined score between both alliances. For the Elimination Matches, the median single alliance score was 40 pts, or 83 for the combined score. Once again, we see a major difference between Qualification and Elimination Matches as you would expect, albeit a bit more than ideal.

Once again, at first glance these match score composition charts are almost identical to what we saw at the last scrimmage. Cones are contributing about 25% of the final scores, and mobile goals accounting for about 50% of the score in the Qualification Matches or nearly 60% in the Elimination Matches. We’re still seeing maybe just one highest stack bonus per alliance per match, and about 1 robot parked per alliance at the end of each game (still about 5-7% of the final score though).

As mobile goal scoring is still quite overwhelming for the final score, it’s nice to see where they’re being scored:

Similar to last time, most mobile goals weren’t score in the Qualification Matches, and of the mobile goals that were scored, the 10pt Zone was by far the most popular for the Elimination Matches. Very little was scored in the 20pt Zone, especially for the Qualification Matches.

Looking into the activity regarding cone stacking, we can see how many cones were stacked, and the height of those stacks:

As we’ve already discussed, not many cones were stacked, supported by the chart on the left. On the right, we see that most stacks were short, between 1-3 cones high, and the highest stack was 9 cones tall. There wasn’t too much difference in stack heights between the Qualification and Elimination Matches.

Other statistics:

  • In only about 24% of Matches was an Autonomous Bonus awarded
  • No attempts at the skills challenges
  • Highest single alliance match score was 89 points
  • Highest combined alliance match score was 150
  • Median difference in Qualification Match Scores was 15 points

As always, if there’s something you’re curious to know we’ll try our best to help, although specific robot design questions won’t be answered.

Thanks to all the volunteers who helped out at this event, and especially to those who recorded all of this data for us to share. The standard Tournament Manager exported data doesn’t give us the information to know the heights of each stacks, so we now have an extra scorer recording all the data separately. A big thank you to them for recording so much information. We have grand plans for what else we can do with the data, although how much of that comes to fruition is a different matter – watch this space!

Also, thank you to Kristin School for letting us use their fantastic venue. The next event will be on the 12th August, at Lynfield College.

9

Was this from 0-0 ties, or were more the likes of 7-7 (single cone scored on stationary goal being the only scoring)?

10

Whilst I have no data to confirm with (we only record the end of match score), it’s safe to say that the clear majority of that 76% was 0-0 scores. There were a lot of robots that didn’t move during autonomous.

11

That was my guess. Thank you!

12

There’s now been two scrimmages since the last update, but since I’ve been unable to find the time in between various assessments at university, I’ve only just managed to get everything done. I’m now using a script to do most of the work for me - previously everything was manually done in a spreadsheet. This is saving me quite a lot of time now. First up, is the third scrimmage (“Scrimmage 4”):

12 August 2017 Auckland Scrimmage Summary

At our fourth In The Zone event of the year, 30 teams showed up – a couple more than last time. Each team was given 6 Qualification Matches before heading into the Alliance Selection, for a total of 45 Qualification Matches.

The rankings heading into the Alliance Selection had 2900A (SymbiOHSis/Onehunga High School) in first, followed by 7682E (Wingus & Dingus) in second and 2941A (Oats Robotics/Otumoetai College) in third. All three of these teams had won all of their Qualification Matches, and were ordered by Autonomous Points. In fact, under the old system, these three would have ranked in the reverse order based on their Strength of Schedule Points.

The Finals were played between the first and second seeded alliances (2900A & 7682E versus 2941A & 2900D). The first seeded alliance won two games in a row with scores of 89-60 and 72-21 to win the event. Congratulations SymbiOHSis and Wingus & Dingus!

On to the number crunching:

http://i.imgur.com/75Cgs1e.png
The big news here is that not only were there no combined scores below 10, there were none below 20 either! That’s an indication that even the lowest scoring matches have improved somewhat, which is great. As usual, we see higher match scores for elimination games than qualification games. The median single alliance score was 28 points for the qualification matches, or 53 for elimination matches. The median combined alliance score was 62 points for qualification matches, or 93 for elimination matches. These numbers are a noticeable improvement from the previous scrimmage.


We see from the match score compositions that there’s been an increase in points almost across the board for placing mobile goals in scoring zones, in particular for the 20-point zone. We see that the mobile goals are definitely the main area of scoring in these games, even more-so than previous scrimmages.


Here we see that more than half of mobile goals are being scored. Particularly noticeable is the improvement with the 20-point zone. Compared to 1% for qualification matches and 8% for elimination matches at the last scrimmage, we’re now looking at 8% for qualification matches and 16% for elimination matches.


As we discovered in the average match score composition charts, the real changes have been in the mobile goal scoring. In fact, comparing the total cones stacked chart to the one from the previous scrimmage, we see what is likely a decrease in cones stacked per game, which is quite unfortunate. The stack heights on mobile goals haven’t changed much, but this is where we get to meet the latest additions to the scrimmage statistics family – a breakdown of stack heights by location:


I’m quite happy with the data in these box and whisker plots. This has been one of my goals from the start, it’s just taken a little longer to bring to reality. The obvious point to notice here is that the tallest stacks are on mobile goals that haven’t been scored in a zone at all. I’ll let you draw your own conclusions from that. Other than that, we’re only typically seeing a few cones stacked on mobile goals, and up to 4 (but not often) on stationary goals. In the elimination matches, it seems that the 10-point zone is a popular place to relocate mobile goals with stacks on them.

Other details:

  • Highest single alliance score of the day: 103
  • Highest combined alliance score of the day: 149
  • Highest stack of the day: 8
  • Qualification matches with an autonomous bonus: 17 (37.8%)
  • Qualification matches with a match affecting autonomous bonus: 1 (2.2%)
  • Median difference in qualification match scores: 32
  • No official robot skills attempts

Thanks to all the volunteers for all of their hard work, and to Lynfield College for hosting this scrimmage.

13

26 August 2017 Auckland Scrimmage Summary

At our fifth event of the season, we had 28 teams come along to Glenfield College. We had 49 Qualification Matches, giving each team 7 games to rank as high as they could before the Alliance Selection.

At the end of the Qualification Matches, the first ranked team was 2900D (SymbiOHSis/Onehunga High School). Coming in at second place were 2918E (GCEC/Glenfield College) followed by 2921V (Free Range Robotics) in third. Just like the last scrimmage, all three of these teams won all of their Qualification Matches.

Our Alliance Selection was interesting in that the top 3 seeded alliances consisted of sister teams – we had 2900D with 2900A in first, 2918E with 2918G in second, and 2921V with 2921X in third. Finals were played between the first and second seeded alliances, just like last time. The first seeded alliance won two games in a row with quite overwhelming scores of 101-26 and 94-28. Congratulations SymbiOHSis/Onehunga High School!

Now for some numbers:


From first glance at the distribution of match scores above, we see something fairly similar to last time – most single alliance qualification match scores being under 30, with a peak in the 50-60 range for the combined alliance scores, and elimination match scores being somewhat higher. For the qualification matches, we saw a median single alliance score of 25 and a median combined alliance score of 53, both a little lower than last time. For the elimination matches, the median single alliance score was 43, and the median for the combined alliance scores at 104 – whilst the former is a little lower than last time, we’ve broken a 100-point barrier with the latter!


Looking into the scores, we see once again that placement of the mobile goals in the scoring zones makes up for the majority of points (about 60-65%). Overall the scores are marginally lower than what we saw from the previous scrimmage, which is what we expected from the median scores discussed earlier.


Once again, we’re seeing statistics that are quite like what happened at the previous scrimmage. For the qualification matches, there’s a bit of a drop in mobile goal scoring overall with a bit of a shift towards the 5-point zone, but nothing significant. For the elimination matches, there’s an overall increase in mobile goal scoring, but again, nothing too significant.


We’re still only seeing a few cones scored per game most of the time, and stacks that are created only being one or two cones tall typically. For the qualification matches, the median number of total cones stacked per game was only four, which isn’t great.


We’re still seeing our tallest stacks in the qualification matches being constructed and abandoned out in the open field, but quite a few are making their way towards the 10-point zone. Interestingly, for the elimination matches, we’re seeing much taller stacks in the 5-point zone to what we saw last time – at the previous scrimmage the only stacks in the 5-point zone for elimination matches were 1 cone tall, so this is a great sign of improvement.

Other details:

  • Highest single alliance score of the day: 105
  • Highest combined alliance score of the day: 149
  • Highest stack of the day: 9
  • Qualification matches with an autonomous bonus: 13 (26.5%)
  • Qualification matches with a match affecting autonomous bonus: 1 (2.0%)
  • Median difference in qualification match scores: 24
  • No official robot skills attempts

Finally, we’ll wrap up with a graph showing how the season has gone so far for us. This graph shows any change in scores of not only the better teams, but also shows the median scores for the losing alliances to try and show some improvement (albeit not much yet) across the board. Of course, everyone is interested in beating the highest score of the day, so that’s included too.

Hopefully, as teams start to develop more robots capable of manipulating cones and mobile goals, we’ll start to see these scores shoot up. With only three more scrimmages until the Asia Pacific Championship in Rotorua in December, I have all of my fingers and toes crossed.

Thanks to my fellow volunteers for helping – with only five of us tertiary students there to run the event we were a little stretched, but managed to do quite a reasonable job of sticking to schedule. Thanks also to Glenfield College for letting us use their venue for this event.

The next scrimmage is on the 9th September at Onehunga High School. I look forward to seeing lots of teams there!

14

@Jij @Welshkiwi Hi guys, on behalf of the limited number of Hawkes bay teams, we were wondering if there are any fields available with game elements? We are trying to get a local vex competition going rather than having teams going to Palmerston north for the day to compete at their scrimmage’s, our robotics mentor said she had ordered a field and game elements as soon as they were released on the kiwibots website but so far nothing has arrived or anything, she has also rung up Chris(i think it was Chris from kiwibots, i can double check if needed) multiple times but his answer has been that they are having a meeting later that week about the issue of fields and other things. It seems however that at least one team in Auckland has got a field with game elements which enables you to hold scrimmages, any information on this situation would be greatly appreciated.
Cheers,
on behalf of Lindisfarne College

15

Hi @optic,

Your best option is to get in touch with Chris Hamling from Kiwibots and see how he can help you out - there’s not much that any of us volunteers can do unfortunately (we only help out because we’re passionate about robotics and STEM, Chris is the man behind Kiwibots). All the fields that were ordered by Chris arrived towards the end of July and have been distributed around the country I believe.

I don’t really know where all the fields are, but I know we have 5 in total in Auckland - all in pretty high demand! That’s spread across upwards of 50 teams (compared to some schools overseas who have multiple fields of their own). I think it’s important that all the regions have at least one field, and ideally as many schools as possible (although obviously with the cost it’s very challenging).

Cheers,
George

16

@Jij
Hi George, any idea when is APAC?
I could be searching for the wrong thing, but some how I couldn’t find it in robotevents.

I am still harbouring a slim hope of sending some teams over this year. Still have great memory of the last APAC in Auckland :slight_smile:

17

Hi @meng, great to hear from you!

APAC is 4th-7th December, in Rotorua. I’ll ask about getting it up on RE soon!

It would be fantastic to see some Singapore teams over here again :slight_smile:

18

@meng
here is the url to the kiwibots information for the event that i could find.

19

9 September 2017 Auckland Scrimmage Summary

We had 28 teams show up to our sixth scrimmage of the year, this time held at Onehunga High School. Every team had 7 Qualification Matches, giving 49 games in total before the Alliance Selection.

SymbiOHSis/Onehunga High School ranked first, second and third after the Qualification Matches. In order of first to third, these were teams 2900D, 2900C, and 2900A. Both the top two ranked teams (2900D and 2900C) won all seven of their games, with 2900D and 2900A winning 50 Autonomous Points across their Qualification Matches, the most scored out of all the teams present.

Finals were played between the first two seeded alliances (2900D & 7682E vs 2900C & 2900A). The first seeded alliance won the first game with a score of 79-49, followed by a second win with a score of 117-50. Congratulations SymbiOHSis and Wingus & Dingus!

Time to jump into the nitty-gritty:


It’s clear when comparing the graphs above with the same from the previous scrimmage, that there’s been an improvement across the board with the match scores. We’re seeing a median of 40 with a maximum of 101 points for a single alliance in a qualification match, and a median of 62 with a maximum of 117 points for a single alliance in an elimination match. For the combined scores (the sum of both alliance scores in a single game) we’re seeing a median of 85 and a maximum of 166 points for a qualification match, or a median of 143 and a maximum of 172 points for an elimination match. These are all a big improvement from last time.


Here we see some improvements from last time. The only minor decreases we’re seeing are with the 5pt zone which makes sense with the transition to scoring in higher scoring zones, and with the parking bonus for the qualification matches (albeit only from 2.2 to 2.0 so not really significant). The big improvements to note are with mobile goals in the 10pt zone and stack bonuses for qualification matches, or cone stacking for the elimination matches. Once again we’re seeing that scoring mobile goals is making up for the majority of points, but with cone stacking and highest stack bonuses becoming increasing more important.


There’s some pretty noticeable changes here compared to last time, especially for the qualification matches. 42% of mobile goals were scored in the 10pt zone for qualification matches, higher than any other scrimmage (previous highest was 24% two scrimmages ago). In elimination matches, more than half of all mobile goals were scored in the 10pt zone (59%, previous highest was 46%). It’s nice to see that most mobile goals are being scored in zones.


For the total cones stacked in matches, we’re seeing some pretty big improvement. We had a maximum of 28 cones stacked in a single game (35% of all cones), which occurred in an elimination match. There was a bit of a drop in the highest stack from 8 to 7, but overall there’s not much change from what we expect for the stack heights just yet.


The only really significant change here is how the stack heights for the 5pt zone have increased for the qualification matches. In elimination matches we’re seeing that the tallest stacks for the 10pt zone and unscored mobile goals have increased from 6 to 7, but the median stack height for the 10pt zone has dropped from 4 to 2.

Other details:

  • Highest single alliance score of the day: 117
  • Highest combined alliance score of the day: 172
  • Highest stack of the day: 7
  • Qualification matches with an autonomous bonus: 18 (36.7%)
  • Qualification matches with a match affecting autonomous bonus: 4 (8.2%)*
  • Median difference in qualification match scores: 30

** This is now calculated slightly differently. For the previous scrimmage, this would have been 12 (24.5%).*

As shown on the final graphs below, we can see that there’s been a big improvement across the board in scores at the sixth scrimmage – not just for the more successful teams, but all teams. The median difference in qualification match scores has increased slightly from 24 to 30, so the gap between teams based on success has increased a little, but it’s great news that the median losing alliance score has increased by the most to date.

Thanks to all volunteers who came along – as you’ll know we can’t run these events without you, especially at the critical times like set up and pack down. Thanks also to Onehunga High School for letting us use their hall.

The next scrimmage is on the 23rd September at The University of Auckland (Newmarket campus). For those 18+ year olds reading this who would also be coming to the scrimmage, remember that this will also be Election Day for New Zealand so please make sure you vote. Of course, early voting opened a couple of days ago, so it’s probably best to vote before the 23rd to avoid any issues with the date coinciding with a scrimmage. It’s not too late to enrol to vote, you can do that at the advance voting place when you vote.

20

Interesting stuff, thanks for sharing.