Autonomous Formula Racing (A2RL) - CLOSER LOOK

[00:00:00]Today we want to take a closer look at a special kind of motor racing and a rapidly growing one, the racing with autonomous cars.

[00:00:08]The most competitive racing series here is the Abu Dhabi Autonomous Racing League, A2RL, which was founded in 2024.

[00:00:18]In its first year, there were eight teams competing, in 2025, already 11.

[00:00:23]In the highest category, Formula cars compete with each other. These are Japanese Super Formula cars with a modified version of the Dallara SF23 chassis, now called the Dallara EAV24.

[00:00:36]But instead of a V8 engine, they have the famous Honda K20C1 engine from the Honda Civic Type R in the back with 550 horsepower and a weight of around 650 kg.

[00:00:49]Each team gets the same car and each car has the same sensor suite on board. GPS devices, left, right, front, rear to locate the car and to check the car's direction and movements.

[00:01:01]Radar, lidar, so a laser scanner to front and sides, and cameras around.

[00:01:08]So, how do you let such a car run on a racetrack as fast as possible?

[00:01:13]First of all, the teams work out an ideal line and speed for the track with countless simulations beforehand, and that's their baseline.

[00:01:22]With this, the car knows how to drive each corner, where to brake, and so on.

[00:01:27]The next step is that the car needs to know where it actually is and in which direction it's going.

[00:01:33]With the GPS signals, the car can find out, but because of obstacles and bridges, etc., GPS can be compromised and you cannot rely on GPS alone.

[00:01:43]So, the teams scan the track before their car drives there and they write software with which they can identify the landscape around the car and work out where it is with cameras and with the lidars.

[00:01:55]And of course, if there's a wall in front of the car, radar, lidar, and cameras will realize that.

[00:02:02]Now, people ask, "What's the difference between radar and lidar?"

[00:02:06]Radar gives you good information, and you can use the Doppler effect to measure relative speeds between objects.

[00:02:12]The lidar, on the other hand, is a lot more precise since the laser is scanning the surroundings and giving you a point cloud of objects.

[00:02:21]Now, the car knows the best line and optimal lap around the track from simulations, and it knows where it is on track.

[00:02:28]You can imagine that all that needs a lot of processing in the background and checking between inputs of different sensors.

[00:02:35]Plus, it's getting increasingly harder to manage all this the faster the car is going.

[00:02:41]You get more vibrations, and if you drive the car on the limit, the car's response is not always the one the software expects.

[00:02:48]And here, the ultimate test is to race the autonomous car against a professional race driver.

[00:02:55]In 2024, the autonomous race car was still 10 seconds slower, which is a lot on a short lap.

[00:03:01]In 2025, the team from Munich was just 1.5 seconds slower than Daniil Kvyat in the same Dallara car, which was an impressive improvement.

[00:03:11]Now, the next much bigger challenge is to race against multiple cars on track.

[00:03:16]So, now you don't have all the space for you, and you don't just have to make sure where you are, you need to monitor where the cars around you are as well.

[00:03:25]Of course, the easiest solution would be to just drive the ideal line and stay behind someone if there is someone in the way. But, the interesting thing now is to write software with some overtaking intelligence.

[00:03:37]So, the software also has a part where it's predicting the behavior of competitors. That's the third part. And the software can also learn. So, if another car is always breaking early at one point, it can adjust to that and maybe use it to overtake in the next lap. That adaption is part four.

[00:03:56]And here, it's quite interesting to see which different approaches the teams are using.

[00:04:01]The two top teams fighting for the win are Uni Morph from Modena, Italy, and the Technical University from Munich, Germany.

[00:04:08]While the Italian team started right at the limit and adjusted, the German team started with a conservative lap and improved lap after lap in the race.

[00:04:17]Also, in terms of braking into the corner, slipstream on the straight, there are many parameters which the teams are adjusting in the behavior of the car. And the really interesting engineering challenge is to talk to professional race drivers about their behavior and style, and to turn this into a software code.

[00:04:36]And the autonomous car has the potential to be faster than humans because they can, for example, brake or four wheels individually, while the driver can just brake off four of them at the same time with a brake pedal.

[00:04:48]So, the development of autonomous race cars is accelerating rapidly, and many students decide for a certain university because they have a team there.

[00:04:58]Again, this is a field where you can learn fast and make quick gains, which you can use for road cars. And since most car manufacturers who are currently working on self-driving cars, there are many career opportunities for team members because they all have real-life experience in this area.

[00:05:15]So, although motorsport is fascinating because it's the connection between man and machine, the development of autonomous race cars is equally fascinating and accelerates the development of self-driving cars.

[00:05:26]Let me know how you like self-driving race cars and the competition with real drivers below, and see you at the next video.