Turning A Super Formula Car Into An Autonomous Racer
Nov 21 2023

Turning a Super Formula Car into an Autonomous Racer

Super Formula produces some of the fastest single-seater racing in the world and is one of Japan’s two biggest professional championships alongside Super GT.

For years, its cars have raced domestically at circuits like Suzuka, Motegi and Fuji, and European viewers have needed to set alarms at absurd times to see them in action. But that will change next year when a field of 10 adapted Super Formula cars, run by university teams, line up for the Abu Dhabi Autonomous Racing League (A2RL). The competition, which will take place at Yas Marina Circuit on April 28th, is aimed at furthering autonomous vehicle technology and raising public awareness through the medium of motorsport.

A2RL is run by ASPIRE, a branch of the Abu Dhabi government’s Advanced Technology Research Council. Big backing is evident, considering the $2.25 million prize pot that has attracted teams from around the world.

And, rather than using the cheaper and easier option of using some decommissioned single seater, ASPIRE went for the Dallara SF23 which only arrived in Super Formula this year. How did it come to pass that such a vehicle ended up in this situation, and how does it differ from the SF23s that can lap under 10 seconds off a Formula 1 car around Suzuka?

According to ASPIRE executive director Tom McCarthy, there were strong links between some of the A2RL project’s technical partners and Italian constructor Dallara, which builds the SF23 chassis. It considered using the Dallara AV-21 (derived from the 2015 Indy Lights car) that has done well in the Indy Autonomous Challenge. But ASPIRE wanted to do something different and go further in terms of base car performance.

After productive discussions with Dallara, ASPIRE’s management went to Japan Racing Promotion, which holds proprietary rights to the SF23, and successfully attained usage of its car for the A2RL competition.

‘There was a huge act of faith from JRP and Dallara to allow us to take their beautiful Super Formula car and start the redesign that would be required for what we’re calling the Emirates Autonomous Vehicle 2024,’ McCarthy tells Racecar Engineering.

ASPIRE then gained other partners, such as Meccanica 42, which developed specific actuators for the autonomous vehicle, and Danisi Engineering which integrated the autonomous stack (consisting of sensors and cameras) into the chassis. Renowned Japanese squad Team LeMans also pitched in for the development process and Yokohama provided its confidential tyres. Brembo supplied the brakes, as it does in Super Formula.

All driverless SF23s will be built specifically for A2RL, rather than being converted from Super Formula spec.

The main difference between the SF23 and the A2RL car is, of course, the absence of a driver and the installation of an autonomous stack in their place. According to McCarthy, the AV apparatus is not much heavier than a cockpit with driver, while the stated total car weight is 20kg heavier than the SF23. The autonomous stack consists of:

• Seven Sony IMX728 cameras with 360deg coverage
• Four AF ProWave RADAR sensors
• Three Innovusion Falcon Kinetic FK1 LIDAR sensors
• A Neousys RGS-8805GC computer

The computer processes the data that the cameras and sensors record. Like any computer, it requires a high amount of cooling, so the car is fitted with an extra set of vents positioned in the nose. The A2RL car also lacks the Halo device that protects the driver’s head in a standard single seater, while the brake lights (unnecessary in an AV) have been replaced with rear-facing cameras. Otherwise, it looks very much like a standard SF23 that races in Japan.

Under the bodywork, the engine is different from what Super Formula teams use. Indianapolis-based 4 Pistons was contracted to develop a race-tuned version of a 2-liter, four-cylinder unit derived from the Honda Civic Type R. The front and rear suspension are both pushrod, like in Super Formula, and have adjustable dampers.

Because the A2RL autonomous stack consists of spec components, the competition is mainly about how teams synthesise, interpret and apply the data that is generated from the cameras and sensors.

Teams will use the same operating system, ROS 2, to work with the data but they will have the freedom to code the cars how they wish using their own software on the side. They are given an autonomous car that can perform all the basic functions, however it’s up to them how to make it go quickly. This relies not only on savvy coding knowledge, but also a sound understanding of how to set up a car mechanically.

‘The drivers are taken out, but the motorsport challenge isn’t,’ McCarthy insists.

‘We’re making sure that it doesn’t come around and there is this genius software programmer who doesn’t know a thing about motorsport and wins. That’s not going to happen.

‘The really demanding stuff for the teams is how well they can construct the element of the coding that we call the planning module. The planning module takes into account what challenges we give them on the track.’

These challenges are yet to be defined although McCarthy has plenty of ideas, from single-car time trials to multi-car races. The latter in particular gives A2RL a unique aspect over other attempts at autonomous motorsport like Roborace and Indy Challenge, although it will inevitably be difficult for many people to resonate with a driverless spectacle.

A2RL track testing is in its early stages. This week the development team was at Dubai Autodrome running a prototype Super Formula car with ex-Formula 1 and current Lamborghini LMDh driver Daniil Kvyat behind the wheel.

This was standard shakedown testing, to sort out the car’s mechanical characteristics and reliability. The plan for next week is to engage a second prototype with the autonomous stack functioning, having built it in Italy and shipped it to the UAE.

Bench testing of the stack has been done already. During the testing process, all of the data is fed back to the teams, enabling them to prepare in advance of getting their hands on the competition cars after winter.

‘We will freeze where we’re at, at the end of December,’ says McCarthy.

‘One of the real purposes of the testing is data collection. Each of the participating teams will be able to train their software on the simulator.

‘They will have the car performance data that we do in testing, and that will be continuously updated. Then, we will do the assembly of the competition cars in February, in the UAE.’

A2RL is assembling a team of mechanics and performance engineers who will be on-hand to teams leading up to and during the competition.

‘From the middle to the end of April, we have exclusive use of Yas Marina Circuit from 8am to midnight,’ adds McCarthy. ‘We have a run plan that each team will go through on each day.

‘The 28th of April will be a culmination of six weeks of team training and two weeks of intense practice.

‘In the best nature of science, it’s about collaboration and competition. In F1, each team is hermetically sealed. Here, every team will be working in a collaborative way.

‘They won’t be giving each other software, but there is a point where you work collaboratively to push the boundary. When you get to race day, that’s where you compete.’

McCarthy hopes that A2RL, with its combination of a Super Formula car and promises of ‘proper’ racing, will be a positive step for autonomous mobility in several ways. There is, of course, the potential for software advancement based on what the teams discover. But he also views the potential for the competition to win over members of the public who are sceptical about AV technology.

‘It brings about greater opportunity for consumers to say this could be safe,’ McCarthy suggests.

‘At the moment, we’re at Level 3 autonomy in cars. People are scared when we put in certain types of assists. They’re turning it off. It feels not as if they’re being helped, but as if they’re losing input. We’ve got to start bringing people along.

‘In the more general sense, people are scared of AI and what it does to them. We’re going to be using AI here, so we need to make sure that consumers are involved in the incorporation so they can see how it can be something that’s positive.’

McCarthy is adamant that ASPIRE isn’t trying to replace motorsport as we know it. Racing is appealing because it puts us fragile humans in control of extremely complex machines. Occasionally, the human loses control, which often makes it even more exciting. That thrill will be lost in autonomous racing, but McCarthy reckons there are other interesting things to gain, such as new engineering questions and answers, and a chance to feed into road-relevant technology.

‘We’re all very clear that nobody is replacing the motorsport of the driver, because who would want to do that?’ he suggests.

‘But what about developing something in parallel that is utilising extreme sports, with science in action?’

Source: racecar-engineering.com