Fraser MasefieldThe Acronis SIT Autonomous team has taken sole ownership of the Season Beta Roborace standings after another polished performance in rounds 4 and 5 at the Las Vegas Motor Speedway.
Coming into the latest race weekend of the world’s first driverless car racing series, Acronis SIT shared the championship lead with rivals Arrival Racing.
Yet, in another weekend punctuated by software problems for other teams, it was Acronis SIT Autonomous who put together two steady rounds as Arrival Racing failed to score any points in round 6, only their second DNF after their disappointment at Bedford Autodrome.
We completed Round 6 of our @roborace – @here Season β 🏁
😩 Arrival and MIT didn't finish.
🏆 Acronis SIT Autonomous has shown excellent results without breaking any obstacle.
💪 The University of Pisa successfully finished the Round and showed how to drift. pic.twitter.com/h2UFSBszhX
— Roborace (@roborace) February 15, 2021
After the wet and rainy conditions that teams had to cope with in the English winter, few problems were expected in Las Vegas. But a freak sandstorm halted proceedings half way through the day before the teams could get running again, and it was Arrival Racing who had the best of it, collecting 42 objects in the Metaverse whilst hitting only three obstacles.
It meant that the British team topped the standings after round 5, Acronis SIT Autonomous also collecting 42 objects but hitting five obstacles. Autonomous Racing Graz were next, collecting 20 objects with 9 obstacles hit, ahead of MIT Driverless and Universita di Pisa, who did not manage to complete their run.
🇮🇹 University of Pisa team
Drift AI lessons 😮 pic.twitter.com/WmuQM7raXV
— Roborace (@roborace) February 14, 2021
It was an incident packed day for Acronis SIT, who had to abandon their own car and plug their software into the standard Devbot 24 used by the other teams. In another change to the format, obstacles hit now means 30 additional seconds are added to team’s overall times, whilst two seconds are deducted for every collectable gained.
“A lot of things happened since our last talk,” said Ilya Shimchik, team lead of SIT Autonomous. “There was a thunderstorm and we had track connection issues so we ran our attempt four times and finally we finished the laps and showed quite good results. But unluckily for us, we hit two additional obstacles on the last lap so we stand in second on the leaderboard.
“We don’t know yet (why the obstacles were hit) because we didn’t see the data set after our lap so right now we’re collecting the data set and trying to analyse. Later we can see. The plan for tomorrow is to analyse the data set, analyse the reasons for hitting the two objects and fine tune our parameters and algorithms so we can show a better result tomorrow. We didn’t speed up our care because in the one test session we had, we faced a significant issue and decided to stay on the safe side.”
Here are the results of Round 5. Go to the thread to catch up on some interesting facts. pic.twitter.com/nCmgkU3qgw
— Roborace (@roborace) February 14, 2021
Whatever Acronis SIT Autonomous did with that data analysis overnight certainly worked in time for round 6 on Sunday, the car and software performing superbly in collecting 30 objects whilst hitting zero obstacles. That placed the team at the top of the leaderboard above Autonomous Racing Graz, who collected just 11 objects but hitting 5 obstacles.
Universita di Pisa had a better day of it, collecting 14 objects but hitting 6 obstacles but crucially for the championship, software issues meant that both Arrival Racing and MIT Driverless failed to complete their runs.
“We had similar problems to yesterday, it was kind of a mechanical failure, which is why we again had to change cars,” explained Shimchik. “After that, we got another problem and our software stopped working. Our first attempt with the next car, we didn’t start and went back to the pit lane. We noticed after a couple of minutes that another software wasn’t running in the background, it was Graz software in the background and our software was mixed with Graz’s software and we couldn’t run.
“After we noticed, we had little time to start and luckily, 20 seconds from the end of our attempt, we started and showed good results, I think.”
There was another worrying moment for SIT Autonomous, as the car appeared to almost grind to a halt on its final lap, before eventually getting going again.
“We had a miscalculation in the algorithm of deviation of obstacles and that is why we decided to increase the safety margin until we find the causes and reasons of these miscalculations. Roborace accepts our calculations or we need to accept Roborace. So, before we resolve this issue, we decided to stay on the safe side with increased safety margins and avoid hitting obstacles.
“It really helps that we’ve already tested this track so I believe for the next round, Round 7, we can increase our performance and push the limits of control of the car.
“Of course I would like to see other teams competing. Arrival of course and MIT I would like to see the full laps but unfortunately they didn’t have a chance to show their performance today. There were also ups and downs for us. I have mixed feeling because we didn’t see other teams such as Arrival and MIT on the track today. Also I am really happy that Pisa and Graz have shown a good result and we see Roborace growing and many teams with good results today.”
Craig Scarborough
