Renault Digital organised last week at Viva Technology the “DIY #Robocars France-Grand Prix Renault Digital”: a unique tournament with autonomous scale-model cars. The objective of the 10 participating teams was to find the best solutions based on artificial intelligence and machine learning technologies to drive the cars themselves as fast as possible around a 69-metre circuit. The challenge was open to employees, professional software coders, engineering students, people from all kinds of backgrounds.
Congratulations to the winning team composed of 4 students from EPITA: Ecole d’Ingénieurs en Informatique, who developed the Alpine 1-Autopylot car in the Alpine colours.
To read the full story: https://bit.ly/3xHglQ9
Robocars : le Grand Prix de voitures autonomes qui a animé VivaTech 2022
L’édition 2022 du salon VivaTech a vu se dérouler un « Grand Prix » pas comme les autres : le challenge « DIY Robocars France », une compétition de mini voitures autonomes organisée par Renault Digital. Au menu, plusieurs types d’épreuves, des concurrents passionnés, un public subjugué et une mise en avant réussie des thématiques de l’intelligence artificielle et du « machine learning ». Reportage en immersion au cœur de cet événement.
There were cars speeding around a track, free trials, hot chases, timed laps and excited fans. So the event on 17 and 18 June at VivaTech was a proper Grand Prix. Just not Formula 1, of course, but it was none the less spectacular. You have to admit that, if you mix electronics, coding, development, artificial intelligence, machine learning* and self-driving scale-model cars, you have a crowd-puller. And that’s what happened on and around the tracks during the two-day event. It didn’t just thrill the audience: the participants thoroughly enjoyed the fun atmosphere, learning opportunity and camaraderie.
Cars learning to drive themselves
Fun, learning and camaraderie was exactly what Renault Digital had in mind when it organised this DIY Robocars France-Grand Prix Renault Digital. The idea came from the United States, and the event attracted 10 or so teams of 2 to 4 people each. The goal was to get 1:10 scale cars to drive themselves around a 69 metre circuit with 2 straight lines and 12 bends as fast as possible. The tournament included several types of events including a timed 2-lap sprint and one-on-one races.
There was also a plentiful variety of pilot. “There were employees, professional software coders, student engineers, people from all sorts of backgrounds competing. That’s what makes Robocars so exciting: seeing so many different angles to reach the same goal: making a car completely autonomous,” says Olivier Bourrassé, Deputy Head of Technology at Renault Digital and the race’s organiser.
The vehicles were off-the-shelf remote-controlled cars, with added-on cameras and sensors. A few months ago, the teams remote-controlled their cars around the circuit to start the training process. Then their job was to develop the best algorithms and tap into artificial intelligence and machine learning to process the data from the sensors and make the cars more and more autonomous over time. Until they knew the circuit by heart, so to speak, and could negotiate it as fast as possible without cutting corners or toppling the cones demarcating it. “A tournament like Robocars is an opportunity for contenders learn more about artificial intelligence, and for spectators and the general public understand it better, or at least demystify it,” Olivier Bourrassé continues.
A friendly atmosphere and helping each other out
They were all shooting for the same target, but each team approached it from a different angle, depending on the challenge it had set itself for the race. They worked on the hardware – using different sensors and cameras – and the technical side. For example, the Alpine 1-Autopylot car developed by Alexandre, Mickaël, Maxime E et Maxime G, four first-year students at EPITA, had larger tyres, so its road grip was tighter and it negotiated bends more effectively, but the team had to remove the car body.
The competition was very friendly and everyone was clearly having a lot of fun. The teams had known each other for several months at that point, and helped one another out more than once. They were constantly chatting via an online messaging app among themselves, with the organisers and with enthusiasts. “The goal for the tournament, more than anything else, was to get all the cars to work properly. We all had trouble with one thing or another. We help each other out a lot; we don’t keep any of our development secrets.” says Cyrille, the “driver” of the Fous du Volant car.
Even the fans helped out. One of the trials called “Catch a Robocar” was a race between an autonomous car and a car that a member of the audience drove with a remote control and immersive vision headset. At least in the event we watched, the human won against the machine.
Having fun, learning and sharing: the contenders unanimously agree that they did all three at this tournament. And more than that: working in the conditions provided an opportunity to expand their technical culture, sharpen their skills and learn from the experience that Renault Digital’s teams have gathered in artificial intelligence and other cutting-edge disciplines.
It was a school project at first, but then a real group of friends started coming together around it. We have learnt a lot about artificial intelligence, from a technical standpoint especially, reports Maxime, from EPITA. While they were fine-tuning their cars, they also started building a network of contacts in the company. Some of the students are now thinking about starting their career at Renault Group.
This kind of playing field for tech enthusiasts is a “magnet” for talent and Renault Group is naturally excited that the students taking part in this tournament are starting to consider their options in the professional world. Renault Group CEO Luca de Meo, who was at the Renault Digital DIY Robocars France stand on Friday, made that point: “We have a very important retraining challenge in-house but we also need to bring in new blood from schools and universities. We need initiatives everywhere that show that changing our culture at Renault also include integrating young people. This is very important to me.”
* A field of artificial technology that can analyse data sets to learn and improve its performance.
Engineers are testing future models long before the first physical prototypes see the light of day thanks to CAVE