We are currently working on other use cases and applications of this model as it has proven highly effective!

This is particularly good as a belt-and-braces approach to ensuring that vehicles are not making poor decisions based on being setup to fail in the real-world physical environment. Our solution ultimately improves the overall safety on roads and motorways.

Applications of this AI model aim at enhancing the security of ADAS and autonomous vehicles by automatically detecting defaced or manipulated signs, sending a signal to the vehicle that the sign should not be trusted.

While trained and tested using a Copper Horse heavily modified mod of a driving simulator game, it proved efficient and accurate on real-world dash cam footage, further confirming the potential of using digital twins to effectively train AI systems.

🚗 Traffic Sign Defacement Recognition:

This model is composed by a YOLO-based object detector interfaced with two Convolutional Neural Networks.

Despite the challenges, Sophia’s determination and skill have been nothing short of inspiring. We couldn’t be prouder to sponsor her journey and can’t wait to see her back on track soon!

Photography: Carl Connikie

resulting in some damage to her Mini. But true to Sophia’s resilience, she’s already gearing up for the next race at Gurston Down Speed Hillclimb next month with repairs underway (new bumper, bonnet, brakes, and radiator incoming!).

It was a challenging day with very wet track conditions, but Sophia didn’t let that stop her—she managed to shave 2 seconds off her time between the first and second practice runs! Unfortunately, Run 3 didn’t go as planned. A brake failure at Esses led to an unfortunate encounter with the Armco...

Have a read of the report and let us know what you think - we have also (as usual), published all of the data openly, licensed under CC BY 4.0. This can be found here: copperhorse.co.uk/resources/

We hope this is useful!