Engineering news
The ‘Robocar’, designed by Roborace chief design officer Daniel Simon, weighs 975kg and measures 4.8m-long and 2m-wide. It has four motors 300kW each, 540kW battery, is predominantly made of carbon fibre and will be capable of speeds of more than 320kph.
The car uses a number of autonomous technologies including five Lidars, two radars, 18 ultrasonic sensors, two optical speed sensors, six AI cameras, GNSS positioning. It is powered by Nvidia’s Drive PX2 brain, capable of up to 24 trillion AI operations per second to be programmed by the racing teams' software engineers using complex algorithms. The brain utilises deep learning for 360-degree situational awareness around the car, to determine precisely where the car is and to compute a safe, efficient trajectory.
The series is designed to be a competition of intelligence so all teams will use the same car, by ensuring the hardware is consistent all efforts will be focused on advancing the software.
“Roborace opens a new dimension where motorsport as we know it meets the unstoppable rise of artificial intelligence. Whilst pushing the boundaries of engineering, we styled every single part of the Robocar,” says Simon.
“We take special pride in revealing a functional machine that stays true to the initial concept shared, a rarity in automotive design and a testament of our determination. It's a great feeling to set this free."
The Roborace Championship is a partnership between Formula E and Kinetik, the London-based futures investment firm founded by Denis Sverdlov. To date, Roborace has been performing demonstrations with its development cars, known as ‘DevBots’, which have a cabin that can be driven by a human or a computer allowing teams to understand how the car thinks and feels on a racetrack alongside real-time data.
In their last outing, two cars raced on a custom-built city street track at Formula E’s ePrix in Buenos Aires. Roborace will continue to use DevBots for demonstrations and testing, but will introduce the Robocar into public displays during the remainder of 2017.
It is hoped the race series will accelerate the development of deep learning systems for safer passenger and commercial vehicles.
“I believe Roborace may be an inspiration to professionals in the automotive, robotics and software domain as the competitive element, and high speeds encourages innovation and fast paced development of faster, more accurate sensors, and faster ways of fusing large volumes of data from several sensor types into meaningful information for the control algorithms to use,” Alexander Eriksson, senior research assistant in the Transportation Research Group at the University of Southampton, tells PE.
“As Roborace offers one team comprised of a crowd-sourced community team, it would offer experts and enthusiasts a great opportunity to engage in the development and de-sign of autonomous vehicle technologies.”
The obvious candidate for the application of the technology developed for Roborace is the automotive sector, he adds, primarily targeting SAE J3016 Level 4 and 5 automation, where accurate sensor fusion with multiple sensors will be a prerequisite.
"This also opens up for the development of deep learning networks for among other things automotive applications. And if raw data from the sensors were to be made available to students or engineers, it may open for learning opportunities and research," he says.
Another likely domain that could benefit could be the development of unmanned aerial vehicles where lightweight sensors, sensor fusion and fast path planning is key for example indoor navigation, adds Eriksson.
The Roboraces are to be scheduled before each Formula E race, in major cities around the world and will be televised in more than 100 countries.