If driverless tech does make its way to F1 is that a prospect we should look forward to? After all, Formula One is defined as much by the adaptive skill of its drivers as the technical specifications of their cars.
While the plan to automate F1 is definitely very much still in development, and we won’t necessarily be seeing driverless version for a few years yet, with companies already working on developing tournaments for autonomous cars, it’s right that we start considering whether or not software alone can really compete with the instincts and reflexes of a real driver, and whether driverless F1 will be something that draws the same crowds anyway?
One of the companies at the forefront of turning F1 driverless is Roborace, developers of the Devbot and Robocar vehicles. In December of last year, they performed the first ever live demonstration of two driverless cars on a track at the same time and they are more broadly looking to develop a racing series for autonomous vehicles.
Featured image courtesy of Roborace. Image courtesy of Roderick Eime
Developing out of Formula E, a racing series using only electric-powered cars, the intention is for teams to compete in the Roborace series using equal cars, each with their own individual real-time computing algorithms and artificial intelligence technologies. Presumably, the inaugural series will act as a sort of proving ground for just what can be done with driverless programming in a race setting.
Bryn Balcombe, CTO Roborace, explained to Factor how he sees the future of the sport. “Formula One will always be the pinnacle of human drivers and powertrain technology. Roborace will always be the pinnacle of AI Drivers and autonomous technology,” says Balcombe. “In fact, Roborace provides a platform to develop vehicle intelligence. In the future that intelligence can be used to make traditional motorsport safer, for example by increasing 360 degree awareness.
“Traditional motorsport is dropping from the mainstream in both audience and awareness whereas Roborace has the ability to reconnect with the public because the same technology will be driving them and their families around in 5 years time. Roborace will introduce competition formats that are exciting to watch, but also represent extreme challenges that would be too risky for human drivers. For example, the introduction of traffic (trucks, buses, vans etc) into the racing environment could introduce closing speeds of over 100mph between vehicles.”
Autonomy versus Instinct
Perhaps the most difficult part of replacing Formula One with a driverless version is that autonomous vehicles, at least in their current form, are unable to replicate human instinct. Their sensor arrays can tell them every detail of the track they are on, and what is around them, but without true artificial intelligence, any reaction must be pre-programmed.
Machine learning will no doubt be able to play a part in teaching the vehicles a vast array of responses and possible choices, but the question remains as to whether or not a car could ever learn enough, and be smart enough, to replicate the split-second decisions that a human makes purely off experience and instinct. Would a car see such decisions too far outside of the norms they are programmed with?
To truly evolve the sport, the vehicles will need to be able to surprise spectators; they’ll need to act in ways that may initially seem counterintuitive
“We refer to the driver of our autonomous vehicles as the AI driver. The AI driver, just like a human driver, uses all of the information available to it from its large sensor suite of Lidar, radar, ultrasonic sensors, GNSS and machine vision cameras, to make decisions about how to drive and react,” says Balcombe. “The primary focus is on the ability of the AI Drivers to perceive and act within the dynamic environments that we create. If an AI Driver is more accurate in perception it has a better chance of taking the correct actions.
“However, actions also require judgement – of intent, of other competitors and prediction. It’s the ground based equivalent of a military dogfight with AI Drivers continually engaged in an OODA Loop (observe, orient, decide, and act). The key challenge is to get inside the OODA loop of a competitor to gain an advantage.”
For now, the question is somewhat moot. The technology isn’t yet where it would need to be to stage a racing series that can compete with the F1 establishment. At the most basic level, it would be relatively easy to programme the vehicles with an ideal driving line and have them follow it. The trouble there is that you would then just end up with a queue of cars that are all unable to adapt and work out a way to get ahead that doesn’t use the line handed to them. To truly evolve the sport, the vehicles will need to be able to surprise spectators; they’ll need to act in ways that may initially seem counterintuitive.
Can we convince people there’s a real driver behind the wheel
In October of 2016, the US Department of Transportation’s National Highway Traffic Safety Administration (NHTSA) updated their defined levels of autonomous driving to reflect the levels outlined by SAE International’s J3016 document. According to the SAE document, at level 5, an autonomous car should be able to equal a human driver in every scenario.
This Level 5 scenario is where driverless vehicles are going to need to be to truly compete with F1, from both a technical and entertainment perspective. For all intents and purposes, a spectator should be convinced that the vehicle is driving as if there was a human behind the wheel. However, the advantage they may hold is their freedom from the somewhat important physical safety of human drivers.
The element of risk involved in sports like F1 is in no small part of the appeal, drawing crowds to witness drivers defying death at high speed. There are of course, however, myriad structures and regulations put in place to try and ensure that those drivers stay as far away from death as possible. Perhaps the greatest strength of a driverless Formula One is that there is no need for such structures because it is only vehicles at risk.
With safety regulations, largely, thrown out of the window is it possible that driverless Formula One could assume a form far greater than the traditional format? Admittedly, providing a car with programming telling it to disregard safety concerns may lead to something closer to rally racing than the perhaps more refined F1. At the same time however, it could prove to be a far more exciting prospect as, free from the frailty of their human masters, the cars are able to work to their full potential. And if it results in a car or two going out in a fireball of glory, it would at least make for a grand spectacle.