Get started Bring yourself up to speed with our introductory content.

AI-driven Robocars are here to burn some serious rubber

Few things can beat the adrenaline rush of being part of a Formula 1 race. The unbeatable concoction of speed, skill and technology makes F1 a spectacular sport. And, not surprisingly, AI is going to make things even more exciting on the racetrack by powering driverless cars.

With the future of passenger cars clearly belonging to driverless cars, it was just a matter of time before someone came up with the idea of driverless racing cars. Enter Roborace, which plans to be the first global championship for electric driverless racing cars. The racing championship will feature 20 cars, 10 teams of two cars each. At the wheel, however, won’t be the superstar F1 drivers, but highly competent AI software smartly controlling all aspects of the race.

What’s inside the Robocar

Based on teardrop aerodynamics and futuristic designs by Daniel Simon, who adds to his credit Hollywood films such as Oblivion and Tron: Legacy, the racing Robocar looks like something straight out of a science-fiction movie. Kinetik, the maker of Robocar, plans to launch four different prototypes. The car features a lightweight carbon fiber body, tires by Michelin and internal computing units by Nvidia.

The power of Robocar comes from a 540 kW battery that powers four motors, 300 kW mounted with each tire. At 4.8 meters long, two meters wide and weighing around 1000 kg, Kinetik expects the car to reach a top speed of 300 km/h (190 mph). The navigation of the Robocar comes from a sophisticated optical system that includes six AI cameras and GNNS positioning, radars, five LIDARs (light detection and ranging) and 18 ultrasonic sensors. Developed in little under a year, the Robocar has an Nvidia Drive PX2 brain that has an open architecture. The AI-driven cars are capable of making 24 trillion operations per second. This entire hardware system takes advantage of deep learning for a 360-degree situational awareness.

The hardware of Robocar is pretty much standard. So, the dominant skill in the race is the programming behind the AI software running these cars.

The future

Robocars can help develop exceptionally safe and efficient technologies for the highways and places where vehicle speed matters. Automobile giants like Ford and Audi are investing large sums of money in driverless technologies. Denis Sverdlov, CEO of Roborace and Charge, wants to create a strong emotional bond between people and AI machines. He foresees that “an emotional connection to driverless cars can bring humans and robots closer together to define our future.”

The challenge

AI for Robocars, like many other platforms, is still in its developmental stages. DevBot is the predecessor AI vehicle of Robocar that can be manned, driverless or both. DevBots were designed from the ground up for the AI development teams to gain firsthand experience in developing software for race vehicles. In a recent Formula E electric car race, a DevBot crashed after miscalculating the corner. It did so while running at a high speed. Such incidents clearly spell out that AI technology for high-speed vehicles still needs a lot of work. Specifically, AI driving these vehicles needs to drastically improve assessment of thousands of factors that come naturally to humans. These include racing conditions, type of surface, how hard and long to brake, taking calculated risk and so on. However, once the technology sufficiently demonstrates that driverless cars are much safer, the adoption of driverless cars will skyrocket.

All IoT Agenda network contributors are responsible for the content and accuracy of their posts. Opinions are of the writers and do not necessarily convey the thoughts of IoT Agenda.