In the early days of racing, it was questionable if a car could even cross the finish line. Vehicle engineering was still a new science, and completing a race in one piece was half the battle.
Now, drivers face off in vehicles optimized for speed -- cars finish fractions of a second from each other, and races are won by a driver's ability to make split-second decisions rather than relying on the design of the vehicle itself. Since cars have become so reliable, some in the sport complain racing has lost some of the original excitement.
"I think motorsports on the whole needs a kick in the ass," said JR Hildebrand, a professional race car driver who has competed in the past six Indianapolis 500s. "We've been doing the same things for a long time."
However, a new kind of race has emerged that echoes the origins of the sport. The cars are much less uniform -- they sport varying arrays of sensors and cameras -- but they do have one thing in common: There is no driver.
While automakers, suppliers and tech companies have taken to public roads to test self-driving cars and train them in real-time traffic, smaller companies and researchers have been pushing this technology to the limit away from pedestrians and other cars, using the racetrack as a playground for autono-mous experiments and inventing new forms of competition along the way.
"The racetrack is a great place to try new stuff out, set it up as a limited environment, no pedestrians, no cross traffic," said Joshua Schachter, an entrepreneur and angel investor who hosts Self Racing Cars, an annual autonomous vehicle event. "It's sort of a special place to be able to do this."
Between races, Hildebrand is helping students at Stanford University develop a self-driving Audi TTS race car -- and it's learning quickly.
"The Audi is already within 1 percent of my lap time," Hildebrand said.
However, he added, the car still isn't capable of learning on the fly as well as humans, a skill some engineers are trying to develop by pushing the technology to the limits in a controlled environment.
In April, Schachter, who also competes in amateur human-driven race leagues with his Spec Miata, hosted the second annual Self Racing Cars at Thunderhill Raceway Park in northern California. Companies such as Comma.ai and AutonomouStuff, and student teams, including one from online classroom Udacity, participated.
The event wasn't a race in the traditional sense. Each team's car did fast laps on the track by itself. Out of the eight teams, just four cars were able to complete a lap. The fastest lap was just under 3 minutes 38 seconds -- a human-driven lap at Thunderhill is typically around 1 minute 45 seconds, according to Schachter.
"It's very casual," Schachter said. "Teams are building and testing interesting things."
Teams have been using the competition as a way to test specific features of their technology and learn from other groups trying different strategies, he said. Like traditional racing, which fostered the development of now common features such as rearview mirrors, innovation is born from the desire to compete.
While watching self-driving cars complete fast laps may seem exciting only to the engineers working on them, some in the field see a potential future for a racing sport involving the new technology.
Roborace is a company that hopes to host autonomous events where entrants race around a track similar to current racing. However, they compete with their own driverless software in identical vehicles. At Nvidia's GPU Technology Conference earlier this month, Roborace CTO Bryn Balcombe said the startup wants to standardize vehicle hardware to ensure the competition is entirely about each competing team's software.
"We're really creating a championship of intelligence," Balcombe said.
Josh Hartung, CEO of PolySync, an autonomous vehicle software development platform, helped organize the first Self Racing Cars event in 2016. Hartung envisions a less conventional future for driverless racing, where spectators can watch self-driving cars break boundaries by completing challenges rather than following the traditional racing format.
"In the golden age of racing, it was a time when dramatic improvements in technology could result in dramatic wins and dramatic competition," Hartung said. "We as humans thrive on that kind of drama."
Hildebrand said traditional racing has become stagnant, with longer races and fewer underdogs. He sees autonomous vehicle racing as a way to break the monotony.
"It needs to find ways to reject the status quo of how racing works," he said, suggesting challenge-based competition rather than a head-to-head format. "Races can really highlight the fact that autonomous vehicles are super capable in a high performance environment."
Like Hartung, Hildebrand said a competition involving both self-driving cars and human-driven cars could prove to be the most exciting.
"That's really man vs. machine," he said. "It's the most interesting relationship in motorsports."