“This technology has increased a lot, a lot in the last couple of years,” says Ducati race boss Gigi Dall’Igna. “We also did many things like this in the past but the results weren’t good, because you need to build up the data. Now AI is really important for us to achieve our results.”
Captain at the helm of the good ship Ducati in Moto GP, Luigi Dall’lgna has steered the factory team to its first world championship in 15 years when Francesco Bagnaia stood on top of the podium in 2022.
Ducati’s recent success didn’t come out of the blue however. This journey started as far back as 2017, when they decided to go “big on data and machine learning to improve MotoGP bikes”. Now, Ducati field 8 bikes in a race making up as much as 36% of the starting grid in Moto GP, and they do this for good reason.
Using AI and Machine Learning means using “Big Data”. You might have seen this term flashed around in the media over the past few years, but what it means is to make the most accurate predictions, we need a lot of data. I wrote an article about this type of thinking in August 2023 called “Making accurate predictions on suspension setup in motocross.” With Ducati having so many bikes on the track, it’s allowing them to create excellent models to test different scenarios and ultimately shave off lap time. For example, in Moto GP, the AI can simulate 8 different line choices in a corner, and determine which is the fastest, giving the rider confidence to know they are making the right decision on track.
This brings us to their Desmo 450 MX motocross project. From the very first pictures of the bike on track, spaterings of sensors have been strewn all over the motorcycle. Ducati were also the only manufacturer to run full data throughout the whole race weekend at the MXGP of Arnhem last month for their debut. In this article, we will take a look at the type of data being measured, and how it could prove useful for racing.
The images above are one’s I’m very proud of. For many months, Ducati have been trusting suspension position measurement to Motoklik sensors. Look closely and you will see our front and rear sensors equipped on the bikes of Cairolli and Lupino. For us as a company, this is very very exciting! To have our products used by a 9x World Champion and such a prestigious brand is a major achievement. I’ve written quite extensively about suspension setup using data, and you can find all the articles on the News page of our website, so I don’t feel the need to go on at length about suspension data here now.
The bottom right picture above shows a satellite position antenna from 2D-Datarecording and a GET rev counter used for having the correct RPM off the start. As the rider looks over the bars at the gate drop, the leds light up when they twist the throttle until they show green, and then the rider knows they are good to go.
As you work your way around the bike, more sensors can be spotted. All 3 shown above are the same type of pressure sensor from Aviorace. By measuring the pressure of the front and rear brake lines, as well as the hydraulic clutch line, you can learn about how the rider rides the bike, and how those behaviours change in different conditions and make adjustments accordingly. For example, you could monitor the precent time a rider has the clutch engaged throughout a lap in different conditions such as sand or hard pack, and use this to decide if a clutch needs to be replaced between motos, or how to change the engine mapping to give the rider more power where they need it.
It can also be used in rider training to see when and how they apply the brakes and if it can be improved. This is maybe a little bit more straightforward in racing on asphalt than in the dirt, but similar principles can be applied. Here’s an example of looking at a “perfect” braking data profile on asphalt.
Finally, we have some sensors more related to mapping and fueling. The lambda sensor on the exhaust is used to measure the amount of oxygen left in the exhaust gasses. Unused oxygen is potentially unused power and you don’t want to leave any of it off the table, especially on the start. It can also be used to adjust the amount of fuel delivered at different altitudes or air pressures as the amount of available oxygen in the air can increase or decrease.
Oil temperature and pressure sensors are used on either side of the crankcases to keep an eye on engine condition. A drop in oil pressure could be a sign that a part has failed inside the motor, while the oil temperature can be used to make sure that there is adequate cooling and the oil isn’t overheating and losing its ability to lubricate and protect engine components.
All of the sensors shown in this article are “extras” from a production model, but further data can be recorded from the ECU such as water temperature, RPM, throttle position, gear position, intake manifold pressure and temperature, mass air flow through the throttle body and so on.
All of this data is only as good as how you can use it though, and therein lies the most important aspect which Dall’Igna alluded to at the start. AI and Machine Learning are enabler technologies which allow an engineer, or end user to turn streams of data from all these sensors into actionable outcomes. A perfect example of this is Motoklik’s AiSetup which transforms suspension data into what to do with the clickers, springs and preload. When you take this a step further in Motorsports applications, the data can be used in full race simulations which can be used to determine which ignition map might be the best, how to trim fueling, how much fuel to put in the bike, and how to valve and spring suspension. Ducati started this in MotoGP back in 2017, so the downside is it takes time to make AI work in Motorsports because of the limited opportunities to gather race data. The thing is though, the sooner you start, the sooner you get there.
No doubt, Ducati will be taking a data driven approach to motocross, as they did in Moto GP, and development of the bike will be rapid as a result.
I hope you enjoyed this article, and you will find many more on our website.
Kind Regards,
CEO & Founder
Jens Köpke