Roborace, the new autonomous racing series that will run alongside Formula E, will surely be a masterful technical exercise that showcases the bleeding edge of modern artificial intelligence technology. Will it be any fun to watch? Given that most of what I love about racing involves a human element, I’m not sold yet.
Most of us have our favorite drivers and personalities. With no driver in the car, the simplest method that many of us use to pick who we’d like to win a sprint race is toast. There is no Kimi, no Junior, or not even a man-like robot named Bender in a driverless racing series. (Maybe someone could put a sticker of Rally Chicken on their car, I guess. I’ll always root for Rally Chicken.)
It’s these very human and not always predictable drivers that make racing worth watching, though. We, as an audience, love to be shocked and surprised. Sometimes drivers take more risks on one lap, then back off for a while to go at maximum attack later. Sometimes they screw up and you’re waiting on pins and needles for them to claw their way back through the field.
No lap is ever perfect, and it’s this quest for perfection—or at least the quest to make fewer mistakes than your rivals—that makes racing a spectacular challenge. We celebrate when people race so well because mistakes are so easy to make, as every lap is burdened with some minor imperfection.
Races are less predictable as a result of their human pilots, and this desire to see a good, close fight draws us back week after week. We even tend to give up on series when they become too formulaic and easy to predict.
Roborace’s competition, on the other hand, will be based around teams’ own algorithms and artificial intelligence. We can’t even pick out a favorite car that’s the most innovative or cool, as they’ll all be exactly alike.
What’s the point in removing the humans from the car but not opening up development on the car itself for teams to build the fastest, most ridiculous creations on four wheels? Clever programming is neat and all, but that isn’t as easy for spectators to follow as cars that are visually different and relying on a variety of innovations to achieve the fastest lap times.
Before I read up on the series’ format, I was hoping Roborace would be an insane engineering war. After all, many of us follow racing for the cutting edge automotive technologies it produces, too. We celebrate creative workarounds of the technical regulations that give teams an edge in competition. There’s nothing more satisfying than finding a loophole to exploit for more speed.
I wanted to see unique cars built to different strengths, especially since restrictions on what the human body can tolerate would be completely irrelevant here.
Nope! Roborace opted for identical spec machinery instead.
Spec series are popular because they give drivers a fairly even playing field so that the most talented racers have a good shot at winning. The chaos of drivers in one big pack of the same cars is racing gold. If you screw up in a race full of identical 911 GT3 Cup cars, for example, you really have to claw tooth and nail back into a competitive position, as you have little to no technical advantage over the other racers. Even the most talented drivers mess up sometimes, producing some legendary charges through the field. After all, they’re only human.
I simply don’t know if Roborace’s A.I. dog-bone cars will produce interesting racing on their own. After all, it’s the unpredictable nature of human drivers, not the perfect-to-the-letter bots, that Google’s autonomous cars have been struggling the most to handle. If every car is plugging away lap after neatly programmed lap, where will the action be? The fastest programmed cars will get out in front, and be able to move around the predictable slower traffic without incident.
You can’t intimidate machinery. You can’t get up in the mirrors of a robot and make them sweat by hanging right off their bumper until they either move over or screw up. Our new robot overlords don’t care that you’re a tough guy. The basic mental games racers play with each other will have no role in this series.
It’s fascinating that Roborace wants to crowd-source one team of experts per race, however, what are they going to do? Talk about what how they programmed their A.I. for the entire race? I guess that’s one way to get around too-predictable action out on track.
If a team’s only advantage in Roborace is in their programming, I’m not sure teams will even share that information with fans at all. Sure, we might be able to deduce a few things from the cars’ behavior, but it will be hard to nail down why cars are behaving differently if teams don’t want to give away their technical advantages.
Will Roborace be a neat technical demonstration of high-tech artificial intelligence capabilities? Absolutely! It’s fascinating to see what we can program now, and given the push for autonomous cars, this kind of thing has actual relevance to cars people want to see more of on the road.
But will it be a good spectator sport? That, I rather doubt. Maybe they need to take a page from BattleBots and allow teams to remotely attack other cars as they make expertly programmed passes.
After all, if there’s nobody inside the cars, at least we won’t feel as bad if they crash.