Vehicle manufacturer Dallara Group and IBM have announced a collaboration aimed at transforming vehicle design through the use of AI and exploring quantum computing.
The project, which combines Dallara’s high-performance automotive engineering with IBM’s physics-based AI expertise, is intended to accelerate aerodynamic design processes and establish more sophisticated workflows for racing and aerospace industries, IBM said in a recent press release.
Dallara is a cornerstone of global motorsport, currently supplying chassis for premier series, including IndyCar. The company’s portfolio also spans Formula 2, Formula 3, Super Formula, and Indy NXT, as well as involvement in Formula E, the World Endurance Championship (WEC), and IMSA.
As part of the project, IBM has been working closely with Dallara to develop domain-specific foundation models that use both Dallara’s high-fidelity aerodynamic simulation data and technical expertise.
According to IBM, both teams plan to integrate validated measurements from real vehicles in wind tunnels and on race tracks in a future phase, but the use of high-quality simulation data alone is already producing compelling early results.
Engineers traditionally rely on computational fluid dynamics to predict aerodynamic forces and optimise vehicle performance across components. These simulations are powerful, but are computationally expensive, with even relatively narrow analyses often taking several hours or more.
Because of this, racecar development workflows may require weeks or months as engineers work through iterations of geometry changes, operating conditions and performance tradeoffs, IBM explained.
The collaboration aims to use AI to accelerate these workflows without replacing the underlying physics.
The traditional approach required a few hours to calculate all the configurations. The AI model completed the same evaluations in approximately 10 seconds, identifying the same optimal design with roughly the same error margins as CFD, IBM noted in its announcement.
This speedup, when applied to a typical complete set of hundreds of geometry configurations, could reduce days of simulation time to minutes.
In a parallel effort, the collaboration is also venturing into quantum and hybrid quantum-classical approaches. The teams are evaluating how these technologies can complement existing simulation methods in the short term while identifying long-term applications for automotive and motorsport design.
Fabrizio Arbucci, CIO of Dallara, said the implications of the project could extend beyond the racetrack.
“More efficient designs could benefit all transport categories, from passenger vehicles to aircraft, and even other industries at the mercy of aerodynamics. Even a 1-2% reduction in drag across passenger vehicles could add up to meaningful fuel-efficiency gains at scale.”
Initial findings from this collaboration were detailed in a study published in April and were recently presented at the International Conference on Learning Representations in Rio de Janeiro.
