CASE STUDY - PREMIER FORMULA 1 TEAM
Virtual sensors: less weight, same visibility.
Predicting slip-sensor signals from the car's other data channels, so physical sensors can come off the car without losing the signal.
Predict slip-sensor signal
3 sensors
20%+ less
Every gram and every wire costs something
Physical sensors on a race car add weight, add points of failure, and add wiring complexity - all against a team that is fighting for every gram of performance. But the signals those sensors capture, like slip, are essential to how engineers understand and tune the car. Removing a sensor without losing its signal would be a meaningful win, but only if the substitute is trustworthy enough to replace it outright.
A sensor built from the other data already on the car
Rather than replacing a sensor with a guess, NEXUS was trained to predict the slip-sensor signal directly from the car's other existing data channels - a virtual sensor standing in for a physical one. The model was evaluated against three separate slip-sensor targets, so the result reflects performance across the sensor set the team actually relies on, not a single favorable case.
A sensor that doesn't need to exist
20%+
reduction in prediction error across three sensor targets
Less weight on the car, same visibility for the engineers: NEXUS's predictions came in over 20% more accurate than the incumbent approach, consistently enough across all three sensors to justify removing the physical hardware.
Weight is performance. Fewer sensors is speed.
In a sport where every gram is fought over, a virtual sensor that reliably replaces a physical one is a direct performance gain - lighter, simpler, and one fewer point of failure, without engineers giving up any of the visibility they depend on.
