How rapid machine learning at the racing edge accelerates Venturi Formula E Team to top-efficiency wins

The next BriefingsDirect Voice of the Customer advanced motorsports efficiency innovation discussion explores some of the edge computing and deep analytics technology behind the Formula E auto racing sport.

Our interview focuses on how data-driven technology and innovation make high-performance electric racing cars an example for all endeavors where limits are continuously tested and bested.

Listen to the podcast. Find it on iTunes. Read a full transcript or download a copy.

<br />

To learn more about the latest in Formula E efficiency strategy and refinement, please welcome Susie Wolff, Team Principal at Venturi Formula E Team in Monaco. The interview is conducted by Dana Gardner, Principal Analyst at Interarbor Solutions.

Here are some excerpts:

Gardner: Aside from providing a great viewing and fan experience, what are the primary motivators for Formula E racing? Why did it start at all?

Wolff: It’s a really interesting story, because Formula E is like a startup. We are only in our fifth season, and Formula E and the management of Formula E disrupted the world of motorsport because it brought to the table a new concept of growing racing.

We race in city centers. That means that the tracks are built up just for one-day events, right in the heart of some of the most iconic capitals throughout the world. Because it’s built up within a city center and it’s usually only a one-day event, you get very limited track time, which is quite unusual in motorsport. In the morning we get up, we test, we go straight into qualifying, and then we race.

Yet, it’s attracting a new audience because people don’t need to travel to a race circuit. They don’t need to buy an expensive ticket. The race comes to the people, as opposed to the people going out to see racing.

Obviously, the technology is something completely new for people. There is very little noise, mostly you hear the whooshing of the cars going past. It’s a showcase for new technologies, which we are all going to see appearing on the road in the next three to five years.

Race down to Electric Avenue

The automotive industry is going through a massive change with electric mobility and motorsport is following suit with Formula E.

We already see some of the applications on the roads, and I think that will increase year on year. What motorsport is so good at is testing and showcasing the very latest technology.

Gardner: I was going to ask you about the noise because I had the privilege and joy of watching a Formula One event in Monaco years ago, and the noise was a big part of it. Aside from these cars being so quiet, what is also different in terms of an electric Formula E race compared to traditional Formula One?

Wolff: The noise is the biggest factor, and that takes a bit of getting used to. It’s the roaring of the engines that creates emotion and passion. Obviously, in the Formula E cars you are missing any kind of noise.

Even the cars we are starting to drive on the roads now have a little electric start and every time I switch it on I think, “Oh, the car is not working, I have a problem.” I forget that there is no noise when you switch an electric car on.

Also, in Formula E, the way that technology is developing and how quickly it’s developing is very clear through the racing. Last season, the drivers had two cars and they had to switch cars in the middle of the race because the battery wouldn’t last long enough for a complete race distance. Now, because the battery technology has advanced so quickly, we are doing one race with one car and one battery. So I think that’s really the beauty of what Formula E is. It’s showcasing this new technology and electric mobility. Add to this the incredible racing and the excitement that brings, and you have a really enticing offering.

Gardner: Please tell us about Venturi, as a startup, and how you became Team Principal. You have been involved with racing for quite some time.

A new way to manage a driving career

Wolff: Yes, my background is predominately in racing. I started racing cars when I was only eight years old, and I made it through the ranks as a racing driver, all the way to becoming a test driver in Formula One.

Then I stepped away and decided to do something completely different and started a second career. I was pretty sure it wouldn’t be in motorsport, because my husband, Toto Wolff, works in motorsport. I didn’t want to work for him and didn’t want to work against him, so I was very much looking for a different challenge and then Venturi came along.

The President of Venturi, a great gentleman, Gildo Pastor, is a pioneer in electric mobility. He was one of the first to see the possibility of using batteries in cars, and he set a number of land speed records — all electric. He joined Formula E from the very beginning, realizing the potential it had.

The team is based in Monaco, which is a very small principality, but one with a very rich history in racing because of the Grand Prix. Gildo had approached me previously when I was still racing to drive for his team in Formula E. I was one of the cynics, not sure Formula E was going to be for the long-term. So I said, “Thank you, but no thank you.”

But then he contacted me last year and said, “Look, I think we should work together. I think you will be fantastic running the team.” We very quickly found a great way to work together, and for me, it was just the perfect challenge. It’s a new form of racing, it’s breaking new ground and it’s at such an exciting stage of development. So, it was the perfect step for me into the business and management side of motorsports.

Gardner: For me, the noise difference is not much of an issue because the geek factorgets me jazzed about automobiles, and I don’t think I am alone in that. I love the technology. I love the idea of the tiny refinements that improve things and that interaction between the best of what people can do and what machines can do.

Tell us about your geek factor. What is new and fascinating for you about Formula E cars? What’s different from the refinement process that goes on with traditional motorsport and the new electric version?

The software challenge

Wolff: It’s a massively different challenge than what we are used to within traditional forms of motorsport.

The new concept behind Formula E has functioned really well. Just this season, for example, we had eight races with eight different winners. In other categories, for example in Formula One, you just don’t get that. There is only the possibility for threeteams to win a race, whereas in Formula E, the competition is very different.

<br />

Also, as a team, we don’t build the cars from scratch. A Formula One team would be responsible for the design and build of their whole car. In Formula E, 80 percent of the car is standardized. So every team receives the same car up to that 80 percent. The last part is the power train, the rear suspension, and some of the rear-end design of the car.

The big challenge within Formula E then, is in the software. It’s ultimately a software race: Who can develop, upgrade, and react quickly enough on the software side. And obviously, as soon as you deal with software, you are dealing with a lot of data.

That’s one of the biggest challenges in Formula E — it’s predominantly a software race as opposed to a hardware race. If it’s hardware, it’s set at the beginning of the season, it’s homologated, and it can’t be changed.

In Formula E, the performance differentiators are the software and how quickly you can analyze, use, and redevelop your data to enable you to find the weak points and correct them quickly enough to bring to the on-track performance.

Gardner: It’s fascinating to me that this could be the ultimate software development challenge, because the 80/20 rule applies to a lot of other software development, too. The first 80 percent can be fairly straightforward and modular; it’s the last 20 percent that can make or break an endeavor.

Tell us about the real-time aspects. Are you refining the software during the race day? How does that possibly happen?

Winning: When preparation meets data

Wolff: Well, the preparation work is a big part of a race performance. We have a simulator based back at our factory in Monaco. That’s where the bulk of the preparation work is done. Because we are dealing with only a one-day event, it means we have to get everything crammed into an eight-hour window, which leaves us very little time between stations to analyze and use the data.

The bulk of the preparation work is done in the simulator back at the factory. Each driver does between four to six days in a simulator preparing for a race. That’s where we do all of the coding and try to find the most efficient ways to get from the start to the finish of the race. That’s where we do the bulk of the analytical work.

When we arrive at the actual race, we are just doing the very fine tweaks because the race day is so compact. It means that you need to be really efficient. You need to minimize the errors and maximize the opportunities, and that’s something that is hugely challenging.

If you had a team of 200 engineers, it would be doable. But in Formula E, the regulations limit you to 20 people on your technical team on a race day. So that means that efficiency is of the utmost importance to get the best performance.

Gardner: I’m sure in the simulation and modeling phase you leverage high-performance computing (HPC) and other data technologies. But I’m particularly curious about that real-time aspect, with a limit of 20 people and the ability to still make some tweaks. How did you solve the data issues in a real-time, intensive, human-factor-limited environment like that?

Wolff: First of all, it’s about getting the right people on-board and being able to work with the right people to make sure that we have the knowhow on the team. The data is real-time, so in a race situation we are aware if there is a problem starting to arise in the car. It’s very much up to the driver to control that themselves, from within the car, because they have a lot of the controls. The very important data numbers are on their steering wheel.

They have the ability to change settings within the car — and that’s also what makes it massively challenging for the driver. This is not just about how fast you can go, it’s also how much extra capacity you have to manage in your car and your battery — to make sure that you are being efficient.

The data is utmost in importance for how it’s created and then how quickly it can be analyzed and used to help performance. That’s something HPE has been a huge benefit for us for. … We can apply that ability to crunch the numbers more quickly.

The data is utmost in importance for how it’s created and then how quickly it can be analyzed and used to help performance. That’s something that Hewlett Packard Enterprise (HPE) has been a huge benefit to us for. First of all, HPE has been able to increase the speed at which we can send data from factory to race track, between engineers. That technology has also increased the level of our simulator and what it’s able to crunch through in the preparation work.

And that was just the start. We are now looking at all the different areas where we can apply that ability to crunch the numbers more quickly. It allows us to look at every different aspect, and it will all come down to those marginal gains in the end.

Gardner: Given that this is a team sport on many levels, you are therefore working with a number of technology partners. What do you look for in a technology partner?

Partner for performance

Wolff: In motorsport, you very quickly realize if you are doing a good job or not. Every second weekend you go racing, and the results are shown on the track. It’s brutal because if you are at the top step of the podium, you have done a great job. If you are at the end, you need to do a better job. That’s a reality check we get every time we go racing.

For us to be the best, we need to work with the best. We’re obviously very keen to always work with the best-in-field, but also with companies able to identify the exact needs we have and build a product or a package that helps us. Within motorsports, it’s very specific. It’s not like a normal IT company or a normal business where you can plug-and-play. We need to redefine what we can do, and what will bring added performance.

We need to work with companies that are agile. Ideally they have experience within motorsports. They know what you need, and they are able to deliver. They know what’s not needed in motorsports because everything is very time sensitive. We need to make sure we are working on the areas that bring performance — and not wasting resources and time in areas that ultimately are not going to help our track performance.

Gardner: A lot of times with motorsports it’s about eking out the most performance and the highest numbers when it comes to variables like skidpad and the amounts of friction versus acceleration. But I can see that Formula E is more about the interplay between the driver, the performance, and the electrical systems efficiency.

Is there something we can learn from Formula E and apply back to the more general electric automobile industry? It seems to me they are also fighting the battle to make the batteries last longest and make the performance so efficient that every electron is used properly.

Wolff: Absolutely. That’s why we have so many manufacturers in Formula E … the biggest names in the industry, like BMW, Audi, Jaguar and now Mercedes and Porsche. They are all in Formula E because they are all using it as a platform to develop and showcase their technology. And there are huge sums of money being spent within the automotive industry now because there is such a race on to get the right technology in the next generation of electric cars. The technology is advancing so quickly. The beauty of Formula E is that we are at the very pinnacle of that.

We are purely performance-based and it means that those race cars and power trains need to be the most efficient, and the quickest. All of the technology and everything that’s learned from the manufacturers doing Formula E eventually filters back into the organizations. It helps them to understand where they can improve and what the main challenges are for their electrification and electric mobility in the end.

Gardner: There is also an auspicious timing element here. You are pursuing the refinement and optimization of electric motorsports at the same time that artificial intelligence (AI) and machine learning (ML) technologies are becoming more pervasive, more accessible, and brought right to the very edge … such as on a steering wheel.

Is there an opportunity for you to also highlight the use of such intelligence technologies? Will data analytics start to infer what should be happening next, rather than just people analyzing data? Is there a new chapter, if you will, in how AI can come to bear on your quest for the Formula E best?

AI accelerates data

Wolff: A new chapter is just beginning. Certainly, in some of the conversations we’ve had with our partners — and particularly with HPE — it’s like opening up a treasure chest, because the one thing we are very good at in motorsports is generating lots of data.

The one thing that we are clear at, and it’s purely down to manpower and time and resource, is the analyzing of data. There is only so much that we have capacity for. And with AI there are a couple of examples that I wouldn’t even want to share because I wouldn’t want my competitors to know what’s possible.

There are a couple of examples where we have seen that AI can constitute the numbers in a matter of seconds and spit out the results. I can’t even comprehend how long it would take us to get to those numbers otherwise. It’s a clear example of how much AI is going to accelerate our learning on the data side, and, particularly, because it’s software, there’s so much analyzing of the data needed to bring new levels of performance. For us it’s going to be game changer and we are only at the start.

It’s incredibly exciting but also so important to make sure that we are getting it right. There is so much possibility that if we don’t get it right, there could be big areas that we could end up losing on.

Gardner: Perhaps soon, race spectators will not only be watching the cars and how fast they are going. Perhaps there will be a dashboard that provides views of the AI environment’s performance, too. It could be a whole new type of viewer experience — when you’re looking at what the AI can do as well as the car. Whoever thought that AI would be a spectator sport?

Wolff: It’s true and it’s not far away. It’s very exciting to think that that could be coming.

<br />

Listen to the podcast. Find it on iTunes. Read a full transcript or download a copy. Sponsor: Hewlett Packard Enterprise.