Through his research, professor of mechanical engineering Chris Gerdes places autonomous cars in one of the most dangerous situations possible: racecar driving.

Much of the excitement surrounding the development of self-driving cars concerns increasing safety and reducing traffic accidents, and autonomous cars often struggle to adapt in different situations, such as winter conditions or construction blockades. Gerdes is investigating the techniques and quick decision-making skills necessary for autonomous vehicles to imitate advanced racecar drivers.

In order to understand how these drivers take advantage of the friction between their tires and the road, Gerdes has been studying the brain activity of expert drivers in action. He has found that expert drivers show little cognitive strain during such measurements and instead rely on muscle memory and instinct.

One challenge of processing the brain wave data is that the waves could be related to thoughts that had nothing to do with driving. Despite these difficulties, however, the data has already been used to create an algorithm for car sliding that imitates the automatic reflexes of experienced drivers when sliding around corners.

Recently, one of Gerdes’ projects, an autonomous Audi TTS known as “Shelley,” beat a human driver around the Thunderhill Raceway Park racetrack for the first time. The human competitor still showed more flexibility than Shelley, however, and briefly drove off the track to gain a competitive advantage.

In the future, Gerdes anticipates a conversation about balancing vehicular efficiency with the need to upload the law. He believes it is possible that the most efficient autonomous car would be one that would deviate from law under specific circumstances.

 

Contact Skylar Cohen at skylarc ‘at’ stanford.edu.