Next year, a squad of upgraded Dallara race cars will hit speeds of up to 200 miles per hour as they race through the legendary Indianapolis Motor Speedway to find out if a computer could be the next Mario Andretti.
The planned Indy Autonomous Challenge, taking place in October 2021 in Indianapolis, is targeting 31 college engineering and computer science teams to push the limits of today’s autonomous car technology. There will be no human racers sitting inside the small cabins of the Dallara IL-15 race cars. Instead, on-board computer systems will take their place, equipped with deep learning software that will allow vehicles to drive themselves.
To win, a team’s autonomous car must be able to complete 20 laps, which equates to just under 50 miles, and cross the finish line first in 25 minutes or less. There is a $ 1 million prize at stake, and the second and third place winners will receive a prize of $ 250,000 and $ 50,000, respectively.
Extreme cases, or outliers, can cause deep learning systems to fail because they haven’t been trained to take them into account.
For example, “fast-moving race cars with obstacles approaching them at ultra-fast speeds,” requiring vehicles to “hold steady and maneuver” across the track, represent an “edge scenario” by excellence, said Peak.
For decades, researchers have tested cutting-edge automotive technologies with high-speed racing vehicles on the circuit for more than 100 years. For example, one race in 1921 tested the capabilities of four-wheel hydraulic brakes in one race, while another race in 1993 tested crash data recorders; Since then, these technologies have become standard in conventional automobiles.
It’s much easier to drive a car to safety if you lose control at speeds of 15 mph, “but it’s catastrophic if you run,” he said.
Currently, a team of researchers at Clemson University is designing specifications for modified Dallara race cars that will be used by all competitors, said Robert Prucka, an associate professor in Clemson’s department of automotive engineering. Competitors are responsible for developing the deep learning systems that the cars will drive.
He said the cars are likely to have powerful engines in the “375 to 400 horsepower range.” For comparison, the average car has a 120 horsepower engine. And instead of one car battery, these race cars will likely have several to overload their computerized guts, Prucka said.
These race cars will also have advanced camera systems, radar technology, and LIDAR technology, which uses lasers to scan a car’s surroundings in order to quickly generate 3D maps, so the vehicle can navigate the roads while avoiding obstacles. While startups and auto companies have generally chosen a particular navigation technology like lidar to power their self-driving cars, these researchers can use the latest systems. Prucka said the event organizers did this because some universities specialize in one navigation technology over the other, and they wanted to create a level playing field for all participants.
Just because competitors have access to the latest lidar and related navigation technology doesn’t mean it’s easy for them to build capable autonomous systems. As Prucka explained, current lidar systems are not designed to scan environments fast enough for extreme speeds, which is a major hurdle.
which produces simulation software for engineers. Although the organizers have not determined what the virtual event will entail, “the intention is that the simulation is close enough to the real race.
If participants manage to complete the simulated race, they will proceed to the physical race. This part of the contest will consist of several days of qualifying races, followed by the big race on October 23.
It is not clear how the current coronavirus pandemic will disrupt plans if the deadly virus remains widespread next fall.
“Right now, we’re going full throttle,” Peak said.
