March ended with the first races of the League and some exciting results. Let’s see who qualified into the finale and who will need to try harder in April. Also, let’s check out who will join the Pro league. We also have the first EMEA Summit winner from Brussels!

All the data for this article has been obtained from the deepracer-race-data repository.

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.

AWS DeepRacer News

Invitation: AWS DeepRacer League March Finale: Monday 11th April 2022

Join the AWS Twitch channel on Monday 11th of April 2022 at 15:00 UTC (Check other time zones here) for some live action where the 16 racers listed below will race for the first three Virtual League spots in the finals.

Invitation: AWS Summits in April

We’re racing in person! Be sure to visit AWS Summits Page and register for one (or more) to compete and qualify into the finals! These are the summits of interest in April:

April races have begun

Visit the AWS Console to join the races in April. We’ve got two great tracks to tackle. Ace Speedway and Ace Super Speedway have been named after the AWS DeepRacer League 2021 runner up JPMC-Ace-Hyderabad. The Open Division track has got a great straight, some quick curves and tricky sharper turns and Pro Division track has taken it to the next level with the same straight but also many mind blowing hairpins and chicanes.

-

Ace Speedwar and Ace Super Speedway (Source: AWS DeepRacer Console)

AWS Summit Brussels – Results

It’s been too long since the last in person summit race in 2019 but now the wait is over! On the 31st of March around 40 racers competed to select the first two participants of the EMEA Summit Online finale: DBro and Chris. You can watch some live action streamed on the AWS DeepRacer Twitch channel.

Pro Division

Rogue Raceway (Source: AWS DeepRacer Console)

Race Info:

  • 76.76 meters long
  • 107 centimetres wide
  • Head-to-bot format with four bot cars cruising at 4 m/s with no lane changes
  • Racers are trying to complete three laps – all laps are separate from standing start
  • Each collision results in a three seconds penalty
  • Each driving off the track results in a three seconds penalty
  • Top 16 go on into a live race which will be streamed on Twitch to fight for places in the League Finals
  • 10 best racers who haven’t yet done so win an AWS DeepRacer EVO
All submissions
Update time: 2022-04-01 01:00:00.000000 UTC

Racers: 82
Total submission attempts: 92413
Best total time: 01:11.254 by JJ
Mean total time: 02:10.338146341
Best lap: 00:23.471 by JJ
Mean best lap: 00:40.680231707
Mean average lap: 00:43.445682927

82 racers is a good prognostic for this season – last year we started in March with 70 Pro racers. With such numbers we should expect more virtual participants in the run up to the finals. If you have been following these updates last year it will be no surprise to you to see JJ at the top of the standings. Again. That said, this was not an easy victory. Let’s have a look at how the top 25 racers did:

Top 25 submissions
Update time: 2022-04-01 01:00:00.000000 UTC

Racers: 25
Total submission attempts: 58805
Best total time: 01:11.254 by JJ
Mean total time: 01:16.982560
Best lap: 00:23.471 by JJ
Mean best lap: 00:25.270200
Mean average lap: 00:25.660480

The differences are relatively small here, and this is what I’ve heard from some of the racers who narrowly missed making it into the top 16 – the competition was fierce and times were improving till the last seconds of the race.

RankRacerTotal TimeAvg LapBest LapResetsSubmissions
1JJ01:11.25400:23.75100:23.47105497
2flatearth01:12.72900:24.24300:23.999056
3rosscomp101:13.52500:24.50800:24.07104287
4AJM-Model-101:14.25100:24.75000:24.52801531
5Rober01:14.26000:24.75300:24.2660504
6DBro01:14.40100:24.80000:24.54402097
7Ensono-Ernesto01:14.88200:24.96000:24.52003581
8TonyJ01:15.13700:25.04500:24.47405514
9Duckworth01:15.45600:25.15200:24.66002137
10PolishThunder01:15.92800:25.30900:24.92806746
11Karl-NAB01:16.32100:25.44000:25.12401153
12AIDeepRacer01:16.77300:25.59100:24.9890450
13ZoD01:16.84200:25.61400:25.06703263
14JPMC-DriftKing-Houston01:17.13800:25.71200:25.6750340
15Deepak-dpk01:17.37800:25.79200:24.93503908
16RobinCastro-DBS01:17.46200:25.82000:25.61003516
17dartjason01:17.92200:25.97400:25.86401046
18Jochem01:18.21600:26.07200:25.6620960
19LeadingAI01:18.65000:26.21600:25.85701877
20Penguin-DNPds01:19.25600:26.41800:25.79701756
21CGS-Huge-Hadr0n01:19.53800:26.51200:26.26703625
22Joffe01:19.91000:26.63600:25.9920978
23DoomBuggy01:20.45800:26.81900:26.66402526
24Aurora-DNPds01:22.87700:27.62500:27.19701343
25ferrite01:2400:2800:27.5940114

The time differences remind my of the most competitive races of the last season and this is in the first month already!

Most of the racers have been around for a long time already and it’s not surprising to see them at the top of the league table. All of them have taken part in at least one finale race last year.

Open Division

Rogue Circuit (Source: AWS DeepRacer Console)

Race Info:

  • 48.07 meters long
  • 107 centimetres wide
  • Time Trial format
  • Racers are trying to complete three laps – all laps are separate from standing start
  • Each driving off the track results in a three seconds penalty
  • Top 10% qualify into the Pro Division and win a Pro Division Welcome Pack
All submissions
Update time: 2022-04-01 01:00:00.000000 UTC

Racers: 1273
Total submission attempts: 18195
Best total time: 00:40.995 by Roster-HTC
Mean total time: 02:23.141399843
Best lap: 00:13.329 by Roster-HTC
Mean best lap: 00:45.925369992
Mean average lap: 00:47.713476826

We’ve had 1273 racers which compared to 972 in 2021 also presents a nice boost from the start. I wonder if we’ll see new all time high – the highest participation to date has been in October 2021 with 2806 racers.

The winner, Roster-HTC, has done a great job and presented very smooth behaviour on track. I wonder if it’s this HTC that they are representing.

1273 racers means 127 new Pro racers in April. Nice! But wait, that’s not all 😀

Top 25 submissions
Update time: 2022-04-01 01:00:00.000000 UTC

Racers: 25
Total submission attempts: 5374
Best total time: 00:40.995 by Roster-HTC
Mean total time: 00:47.755
Best lap: 00:13.329 by Roster-HTC
Mean best lap: 00:15.513040
Mean average lap: 00:15.918

In the top 25 we can see some close times. More interestingly, top 25 racers represent almost 30% of the submission count of the whole group. This makes me suspect there has been a rather active group towards the top of the division.

I decided to dig in a little and I found JJ and Duckworth in that list. That’s right: We have some unexpected participants in the Open Division data! Now that I found it I remembered that there was a little tiny bit of a configuration problem and Pro racers were able to submit to Open Division and not to Pro for a brief moment. They have been reintroduced to Pro but their entries remained in the records. This means that it’s not quite clear to me who qualifies into the Pro division. I’ll try and make a guess by removing them from the list and trying again.

I use Pandas to handle the race data. I have a standings Dataframe for pro: pro_df and for open: open_df. The below allows me to get only Open racers:

x = open_df.append(pro_df).drop_duplicates('UserId', keep='last')
x = x=='OPEN']

With that I got 1236 racers which should mean 123 new Pro racers. Let’s see them (just note that the positions in the list below have not been fixed and represent values from the Dataframe where the Pro racers were still present):

RankRacerTotal TimeAvg LapBest LapResetsSubmissions
1Roster-HTC00:40.99500:13.66500:13.32901839
2Benny-LBS00:40.99600:13.66500:13.40901562
3Boltronic00:41.72200:13.90700:13.805036
4ADA00:42.04600:14.01500:13.5350689
6TestableArchitect#889200:44.66200:14.88700:14.608026
8Bandwidth-RnD00:46.44300:15.48100:15.1850318
9leo-DNPds00:47.29100:15.76300:15.274020
10sneaky-snake00:47.59100:15.86300:15.533042
11NeilSimon00:48.39600:16.13200:15.6080101
12iliu00:48.79700:16.26500:16.201072
13Langley00:48.99600:16.33200:15.995039
14IcePandaBoom00:49.05600:16.35200:15.6040141
15Velocity-Theta00:49.25700:16.41900:15.8720145
17wl00:49.85300:16.61700:16.264055
18ZOOM00:50.53100:16.84300:16.211070
19JaguarRobotics-Blaze00:50.72600:16.90800:16.731028
20Jianlong00:50.78700:16.92900:16.739097
21Groovy-DNPds00:50.86300:16.95400:16.601015
22Technopro#Fujishima00:51.13000:17.04300:16.86304
23Velocity-Eta00:51.38700:17.12900:16.864020
24BlazingPhoenix00:51.66200:17.22000:16.79904
25scudz00:51.72500:17.24100:17.002030
26ParametricNumber#553000:51.79000:17.26300:16.7300141
27Gan00:51.94500:17.31500:16.930031
28tranvietdung-DBS00:51.97700:17.32500:16.659014
29AlexanderJones#100:51.99300:17.33100:16.403010
30AstaRomeo00:52.05300:17.35100:16.059013
31mrRobot00:52.12300:17.37400:17.133023
33Velocity-Delta00:52.52300:17.50700:17.330017
35technopro#sakamakikazumi00:52.91700:17.63900:17.196020
37BlockchainNotation#367800:53.11700:17.70500:17.198022
38muzi00:53.45900:17.81900:15.727259
39Shreyasi00:54.19500:18.06500:17.398013
40ocroucher00:54.19700:18.06500:17.871058
41AccessibleScript#148600:54.45900:18.15300:17.800063
42PaloAlto#109500:54.65700:18.21900:17.997013
43IVECloud-200:55.03200:18.34400:14.998216
44MartinoC00:55.12600:18.37500:18.143040
45asache300:55.32400:18.44100:18.13402
47mumoody00:55.65000:18.55000:18.003019
48TigPT-tecRacer00:56.12800:18.70900:17.266123
50DannyS00:56.58400:18.86100:18.59403
51livDaliberti00:56.59400:18.86400:17.6640142
54Baris00:56.72900:18.90900:18.33604
55F1-DevelopersIO00:56.78800:18.92900:18.1450539
56MachineExpression#213500:57.11800:19.03900:18.728010
57ish00:57.25600:19.08500:18.66507
58Zelenskyy00:57.45200:19.15000:18.931015
59arcat00:57.72900:19.24300:18.929045
60kimmizian00:57.80600:19.26800:16.52927
61ManchesterW00:57.86000:19.28600:19.067017
62JRF00:58.78500:19.59500:19.459011
63BetaNetwork#783000:58.78900:19.59600:17.26924
64FuturisticAttention#606700:58.87700:19.62500:19.065018
65KORJJM00:58.92300:19.64100:19.322014
66FrostyVT00:59.12800:19.70900:18.40519
67TheSecondBit00:59.12800:19.70900:19.278014
68IVECLoud00:59.20000:19.73300:19.667014
69DeposeE00:59.32400:19.77400:19.53203
70NumericTheory#276600:59.66200:19.88700:17.670117
71Mukundan-AMPS-0400:59.92200:19.97400:19.66707
72Racer-Matic00:59.98000:19.99300:19.46405
73Pyrenees01:00.52900:20.17600:19.94908
74Aj#432101:00.59500:20.19800:18.799139
75Cinderpelt01:00.99000:20.33000:19.129127
76husky0101:0100:20.33300:19.268023
77UberAllen01:01.13000:20.37600:17.9832335
78GT-DevelopersIO01:01.18800:20.39600:18.26522
79Matt01:01.21500:20.40500:18.87018
80BlockchainComment#843301:01.25700:20.41900:20.134010
813s-Champion01:01.32300:20.44100:19.997013
82webdiver01:01.46200:20.48700:18.93314
83Ashiq01:01.71800:20.57200:19.798016
84GenerativePackage#922401:01.85400:20.61800:20.33703
85ProteusXplorer01:02.06600:20.68800:19.5300120
86Jasper01:02.19300:20.73100:19.259121
87DRD-181201:02.32300:20.77400:20.464033
88robracer01:02.46400:20.82100:20.06404
89Skyline01:02.52500:20.84100:18.657216
90AtomicObject#605301:02.59900:20.86600:20.607013
91Heath01:02.65500:20.88500:20.33205
92JaniSeppala01:02.66100:20.88700:19.25523
93NFS-DevelopersIO01:02.72500:20.90800:20.59706
94SlowAndUnsteady01:02.72700:20.90900:20.596317
95QUICK-DeepRacer01:02.85900:20.95300:20.59709
96Ferahgo01:02.86000:20.95300:20.735011
97Suzy01:02.92800:20.97600:19.402112
98Deployment-Sydney01:03.25400:21.08400:19.52842
99JaguarRobotics-Turbo01:03.31900:21.10600:20.799020
100ImmutableNumber#166201:03.38600:21.12800:20.728031
101USYS-infra-player01:03.39400:21.13100:20.603010
102FocusedDelta#323001:03.53500:21.17800:19.602121
104nine901:03.63000:21.21000:20.666015
105WeiJian-SP2P01:03.79200:21.26400:20.86904
106BooleanObject#317601:03.87300:21.29100:20.320127
107TestableNumber#630301:04.05800:21.35200:20.99809
108bob12401:04.33800:21.44600:21.13435
109Maximus-KK01:04.45700:21.48500:21.32806
110FTM-41401:04.58500:21.52800:21.260041
111Ethan-weber01:05.32400:21.77400:18.734210
112FlosRacers01:05.32700:21.77500:20.32918
113PassingSwitch#470201:05.59500:21.86500:21.734010
114KDE01:05.79600:21.93200:21.66805
1150-dayAttack01:05.91800:21.97200:20.867022
116niro01:05.92800:21.97600:20.666415
117CriticalSystem#909101:06.19700:22.06500:19.394238
118oldmon01:06.46000:22.15300:21.66036
119RyushoYosei01:06.53300:22.17700:21.80208
120Fungible01:06.59300:22.19700:21.79507
121babu#336401:06.72600:22.24200:22.005091
122Dvorson01:06.72900:22.24300:21.46504
123Adam-Niu01:06.85600:22.28500:18.995226
124SolvableServer#138801:06.92800:22.30900:20.803127
125INFOXNET-KMS-DeepRacer-04245401:06.94300:22.31400:19.39725
126Dragonscelt-cwagner01:06.99400:22.33100:21.99302
127onepau01:07.25200:22.41700:20.871113
128TabletConsole#762401:07.31400:22.43800:20.050218
129Kyndryl-Solution-Engineering01:07.32300:22.44100:22.26009
131RajanSingh01:07.55700:22.51900:20.80318
132Ghent01:07.65800:22.55200:22.195017
133IcemanTaylor01:07.66100:22.55300:21.677035
134UnicodeCircuit#366201:07.79400:22.59800:19.796216
135NumericLambda#667001:07.98000:22.66000:19.13153

NumericLambda#6670 has (according to my guess on how AWS have interpreted the current situation) made it into the Pro Division as the last person in March, with a time of 01:07.980.

A quick glance over the list has drawn my attention to a few of know racers who returned to racing in March: Boltronic, ADA, Bandwidth-RnD (who has published a great article on racing recently), leo-DNPds, IcePandaBoom to name a few. Welcome back, friends! And to everyone else in this list: Welcome and let’s make friends in the Community Discord!