AWS DeepRacer League 2021 Update #13

Join me as I look at the first two weeks of racing in May’s AWS DeepRacer League.

AWS DeepRacer

AWS DeepRacer is a 1/18th scale autonomous race car but also much more. It is a complete program that has helped thousands of employees in numerous organizations begin their educational journey into machine learning through fun and rivalry.

Visit AWS DeepRacer page to learn more about how it can help you and your organization begin and progress the journey towards machine learning.

Join the AWS Machine Learning Community to talk to people who have used DeepRacer in their learning experience.

DeepRacer News

DeepRacer Open-Sourced

AWS DeepRacer car software is now open source. This means you can learn more than before about how the car software is working but not only that: you can also modify it to build your own project on top of the device, not just race! You can see the code in DeepRacer repo on GitHub.

On top of all that you can also purchase new accessories for your car. Sample projects include works by Community’s Chris Miller, Jochem Lugtenburg and Martin Paradesi. Read more about them on the AWS DeepRacer Robotics Project Page.

DeepRacer Underground

If you’ve missed the last AWS DeepRacer Underground, go to Twitch to view the replay.

Pro Division

Kuei Super Raceway (Source: AWS DeepRacer Console)

Race info:

  • Track named after Kuei-NCTU-CGI from Taiwan who finished third in 2021 DeepRacer Championship
  • 60.74 meters long
  • 107 centimeters wide
  • Head-to-bot format with four bot cars cruising at 1 m/s without lane changes
  • Racers are trying to complete three laps
  • Each collision and each driving off the track results in a five seconds penalty
  • Top 16 go on into a live race which will be streamed on Twitch to fight for places in the League Finals and AWS DeepRacer EVOs

Let’s start with the general summary:

All submissions
Update time: 2021-05-17 21:00:00 UTC

Racers: 105
Total submission attempts: 18820
Best total time: 01:04.678 by JJ
Mean total time: 03:51.427666667
Best lap: 00:21.080 by JJ
Mean best lap: 01:06.850771429
Mean average lap: 01:17.142247619

Unsurprisingly JJ is leading the way. We’re already at over 100 racers which makes sense taken into account that we are nearing 300 Pro racers. It’s a bit more interesting in the top 20:

Top 20 submissions
Update time: 2021-05-17 21:00:00 UTC

Racers: 20
Total submission attempts: 16975
Best total time: 01:04.678 by JJ
Mean total time: 01:43.697800
Best lap: 00:21.080 by JJ
Mean best lap: 00:31.942900
Mean average lap: 00:34.565650

The mean total time is quite behind JJ. I’m actually quite surprised that there still are times over two minutes in top 20. We’ve got three finalists in the top 20 so it still shows everyone who is in the finale zone, but if they keep racing I might need to expand to top 30 next time.

Let’s see the top 20:

Rank Racer Total Time Avg Lap Best Lap Resets Submissions
1 JJ 01:04.678 00:21.559 00:21.080 0 1185
2 Karl-NAB 01:04.990 00:21.663 00:21.310 0 1792
3 dartjason 01:14.817 00:24.939 00:24.529 0 761
4 rosscomp1 01:16.213 00:25.404 00:24.179 0 864
5 Duckworth 01:17.497 00:25.832 00:25.006 0 316
6 Ernesto 01:18.804 00:26.268 00:24.982 0 165
7 Jochem 01:20.221 00:26.740 00:24.491 1 8
8 Matt 01:23.362 00:27.787 00:27.196 0 724
9 alexj 01:40.634 00:33.544 00:32.424 0 28
10 PolishThunder 01:49.222 00:36.407 00:30.706 5 2652
11 SorinB 01:52.120 00:37.373 00:35.092 1 1616
12 kimwooglae 02:00.753 00:40.251 00:37.243 7 2459
13 ep-dr-racer1 02:02.557 00:40.852 00:34.039 7 774
14 kimmizian 02:03.496 00:41.165 00:38.536 4 1380
15 think 02:03.960 00:41.320 00:37.170 8 1328
16 blueberrypie 02:09.456 00:43.152 00:40.313 8 799
17 GregRacing 02:09.493 00:43.164 00:38.190 5 20
18 drunken-monkey 02:09.700 00:43.233 00:40.900 1 6
19 kim 02:13.432 00:44.477 00:43.931 3 25
20 ProteusXplorer 02:18.551 00:46.183 00:37.541 4 73

Currently we’re still missing GT-DevelopersIO who raced wonderfully in the April’s finale and lost his qualification in the last runs of the race. I’m sure we’ll see him again. NYCU students have not made an appearance just yet. They have made their last day entries a signature move, but then in April it left them all in places 20-40. Will they manage to improve this month?

Finally, it is important to mention two rookies: GregRacing and kim have started racing in 2021, qualified into the Pro Division and now are in the finale spots. They will however have to up their game a bit to increase their chances as they are in the hot seats, joined by a returner drunken-monkey who has been in the community since 2019 but has had a longer break. Good to see him back not just on Slack.

Open Division

Kuei Raceway (Source: AWS DeepRacer Console)

Race info:

  • 45.193 meters long
  • 107 centimetres wide
  • Simpler and shorter than the pro track but still pretty challenging with a variety of straights and turns
  • Time Trial format
  • Competitors need to complete three laps and total time counts for their classification
  • Each off-track results in five seconds penalty.
  • Racers fight for qualification to the Pro group which is won by the top 10%
  • They will also receive a Pro Welcoming Kit which is a bit of a mystery

Let’s start with overall stats:

All submissions
Update time: 2021-05-17 21:00:00 UTC

Racers: 354
Total submission attempts: 2458
Best total time: 01:23.839 by Otaboroid
Mean total time: 04:38.314717514
Best lap: 00:27.078 by svenus
Mean best lap: 01:20.092282486
Mean average lap: 01:32.771251412

354 racers sounds like a bit of a slow-down in new appearances. New waves usually show up around the events just like the last month’s India Women’s League. This month I’m expecting to see a few groups when the AWS Summit Online pick up. They usually have some workshops and races so there’s often a chance to get some credits (bear in mind I haven’t checked if there are any credits giveaways).

So far this means 35 racers to qualify into the Pro Division.

Top 20 submissions
Update time: 2021-05-17 21:00:00 UTC

Racers: 20
Total submission attempts: 477
Best total time: 01:23.839 by Otaboroid
Mean total time: 02:02.231650
Best lap: 00:27.078 by svenus
Mean best lap: 00:36.637250
Mean average lap: 00:40.743650

Again, we have the split between the best total time and the best lap time. Otaboroid from Nigeria has started racing in October 2020, in the reInvent Wildcard race he ended 17th which was pretty close to a qualification. This month it looks like this is a month to qualify into the Pro Division. Svenus has been in the finals in 2020 and is now in the second, with a pretty high chance to qualify as well. Let’s have a look at the whole top 20:

Rank Racer Total Time Avg Lap Best Lap Resets Submissions
1 Otaboroid 01:23.839 00:27.946 00:27.535 0 108
2 svenus 01:26.118 00:28.706 00:27.078 1 54
3 mz-sin 01:29.450 00:29.816 00:29.722 0 25
4 edxtore 01:39.776 00:33.258 00:33.182 0 3
5 Dzmitry 01:42.717 00:34.239 00:30.816 5 58
6 Polaris-DNPds 01:58.540 00:39.513 00:38.203 3 3
7 BrilliantDragon 01:58.584 00:39.528 00:34.399 3 1
8 mansoor 02:01.525 00:40.508 00:30.999 7 37
9 phu 02:02.011 00:40.670 00:36.455 2 24
10 THE-REAL-GODZILLA 02:02.244 00:40.748 00:37.324 2 41
11 Aurora-DNPds 02:07.822 00:42.607 00:36.156 6 6
12 Hyderabad-3-Vega 02:09.454 00:43.151 00:40.675 1 4
13 IceManNCSU 02:15.246 00:45.082 00:35.469 7 22
14 basha 02:15.881 00:45.293 00:36.869 7 4
15 cyyyrus 02:19.680 00:46.560 00:41.075 4 29
16 T-Matsu 02:20.998 00:46.999 00:41.553 6 5
17 DataVroom 02:21.601 00:47.200 00:45.086 7 31
18 Talladega-Nights 02:21.933 00:47.311 00:45.329 10 8
19 ABRacer 02:23.373 00:47.791 00:41.813 12 2
20 CopenhagenRacer123 02:23.841 00:47.947 00:43.007 9 12

In the top 20 we’ve got quite a company: Hyderabad-3-Vega has been a finalist through the Wildacrd racer, IceManNCSU has been around since 2019, we’ve got two DNPds racers from Japan. Quite interesting racers to follow in the weeks to come.

Remember to join us next week to see how the races unfold!

Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.