Italian super bike maker and world leader Ducati has been nailing back-to-back Moto GP victories and Accenture Analytics has a big role to play in the wins. The Bologna headquartered Italian company turned to artificial intelligence and machine learning to crank out simulated test data.
Analytics India Magazine spoke to Marco Vernocchi, EALA Lead at Accenture Analytics, part of Accenture Digital who gave us the lowdown on how the Ducati Corse team arrives at best decisions from its training data faster and more effectively. The Ducati-Accenture partnership goes back to 2012 and now continuing onto the race track. Of late, Accenture Analytics is enabling the Ducati team to create an intelligent testing approach with a bespoke analytics engine.
To date, around 4,000 sectors of race tracks and more than 30 different racing scenarios have been analysed, with a wider roll-out of the solution expected. “In testing, data such as engine running parameters, speed, revs and tyre and brake temperatures are collected, and the Ducati Team will use this to plan, prepare and test for MotoGP races,” said Vernocchi.
MotoGP races being won on the back of Big Data
Behind the race track performances is data gleaned from 100 IoT sensors on the bikes and existing testing data. The Accenture solution allows team engineers to create new perspectives by simulating and assessing bike performance under a range of conditions. By applying advanced analytics and machine learning techniques, simulated results based on data from previous tests enable engineers to optimize bike configurations for any MotoGP race.
“To put things in perspective, there are 18 MotoGP tracks, and the Ducati Corse team needs to test as many configurations and scenarios as possible for each of these tracks to make sure the bikes perform to their limit,” shared Vernocchi.
We can then see the impact of these settings, and make predictions about how changing one setting or another might impact the performance of a bike on the test track. This enables more potential configurations than ever before to be tested, maximizing the benefits of on-track testing and gaining competitive advantage for race day.
How Ducati Corse is creating bikes that become smarter with every test run:
- Sensors on the bikes gather data, which is then captured before the Accenture Analytics algorithm is applied to it
- Test engineers are using real insights to alter bike configurations depending on the track conditions, adding new value to on-track, intelligent testing
- By simulating and monitoring a motorbike’s performance under a vast array of track and weather conditions Ducati Corse is applying ML techniques integrated with IoT sensor data saving time, expense and effort of traditional on-track testing
- Data visualization tools designed for an intuitive user enable testing engineers to interact with insights and find new perspective on configurations and race times
In the world of racing, making big gains from big data is nothing new. Leading motorcycle manufacturers – Yamaha, Honda are are plowing in massive investment to narrow the performance gap and clinch podium finishes. Honda debuted a self-balancing bike that can even ride itself at CES in Las Vegas. The global debut at CES saw Honda unfurling its Moto Riding Assist technology, which leverages Honda’s robotics technology to create a self-balancing motorcycle that greatly reduces the possibility of falling over while the motorcycle is at rest.
Winning big with Integrated Machine Learning technology
Vernocchi believes that with integrated Machine Learning technologies, there is more data flowing into the system – which means there are more configurations available for testing with increasingly accurate performance predictions. Have the results been implemented on live tracks?
“At this point we are running a pilot project, but so far, we’ve seen excellent results in the lab with the Accenture solution. The ability to use existing and new testing data will help Ducati choose the optimal configuration for its bikes,” he said.
Try deep learning using MATLAB