Is Roborace the Future of Racing?

[phantomvox]

In my opinion what makes racing exciting is the driver. Watching Gasly struggling driving a Red Bull car, kimi finishing top ten in an Alfa Romeo, Grosjean crashing constantly or Sainz singing smooth operator are stuff that self driving cars cant replicate.

[HomieFFM]

Single Player racing sessions with AI in real life... will be possible.... if you are rich enough xD

[STF_YVR]

There is your average "watch me race" sim racing youtuber and there is Chris Haye who produces original, researched and interesting content every time. Thank You Chris.

[willbennett]

Bro itll be people driving with a thrustmaster with cemeras in the car it will be real life sim racing

[gseric]

I couldn't even imagine the effort it would take into coding a program to emulate the steering/pedal/braking inputs a professional driver has to take, and just on one particular track. A track is always evolutionizing at small increments as well, so the AI has to be able to learn these small changes at a whim. That sounds truly challenging...

[miguelamorim]

Hello Chris and congratulations on your work.
About Roborace; well, if it is indeed the future, then the SimRacers of the present sure have promising times (jobs) ahead :-)

[sambaraka]

Esport is the future of all sport. Having race tracks, stadiums, travelling to said places doesn't make sense in a vr future. Bits > Atoms.
Before anyone gets romantic about "the real thing", letters, browsing music stores, face to face dating and live theatre were "real things" too.
Also Esport motorsport could bring in/be targeted at a whole new audience same way horse race fans didn't necessarily convert to motorsport fans.
But in 2050 the largest competitions will all be digital.

[giuniral]

What is fascinating about this concept is that the engineers, software engineers and other people from "behind the scenes" will be brough to the spotlight... in most of the motorsports they're usually behind the stage.

This video looks like The Drive or Drivetirbe content. Great job.

[zerobandwidth]

This also opens up the possibility of a really interesting pro/am series, where half the stints are completed by the AI driver, and the other half by sim racers strapped into a rig in the paddock, connected remotely to the real car, and using the car's own visual sensors for feedback. (OK, so, essentially, drone racing.) All the tech of RoboRace, all the fandom of human drivers, and none of the danger of smashing your human drivers into a catch fence at 300kph.

[lucaruocco]

Congratulations for the work Chris, you touched a topic which is really interesting. But I ask myself, if we take the human factor out of racing, what remains? The wonderful thing of racing is how a human can take a vehicle, which ever it could be, beyond its limits. That's the real emotion, what connects us to motorsport, creating a bond between a man driving faster and faster, improving every corner, every lap, and us normal people, trying to do the same in our life. A code written by someone behind a PC can be fascinating but it will never, never reach the real magic of motorsport

[schmidtvonhallzig]

Nice, fresh real world video approach. Thank you! I saw the Roborace Car at the FOS Broadcast. It is interesting, but competitionwise it will be a niche for a long time. Such as robot football. Even in the long run, I don't see too much potential for a fully autonomous racing league. Especially with the Open Source approach it will be a good carrier for development of autonomous driving.

[Ben-qutq]

I personally am very very hyped for this. In the future, when the machines almost inevitably outperform us and even the best drivers, we'll be able to learn things from being able to watch the robots driving.

[vatch]

This is amazing. If they make this work well, it opens up so much possibility on and off track. Great video, and an awesome topic, Chris! Cheers!

[kanojo]

Probably the most interesting thing about this right now is that NO major car manufacturers are involved. That's weird... they are ALL doing significant autonomous driving research right now, you'd think this would be an automatic opt-in for Mercedes, or Google, or Uber, or Tesla... or any one of a dozen others.


I wonder why that is.

[Ofitus]

This is undoubtedly an unbelievable feat of technology and ingenuity. The fact this is even a thing is incredible, and we should recognise the people at Roborace for pulling this off. But I struggle to see the appeal as a racing series. Watching a race is exciting because of the drivers inside the cars pushing to the limit, giving it everything to win. A battle of 2 AI cars won't put us at the edge of our seats like a battle between Leclerc and Verstappen does for example. And as much as the team development will be interesting to watch, it'll miss those on track battles that make us love this sport

[jackrowe]

Great vid Chris!👍 Robo race will come the future and we are just waiting for the series to come out. If robo race takes over I think the racing atmosphere will slowly disappear.

[pedrospeeder]

Huge wall of text incoming... As a mechanical engineering student, I have... Mixed feelings about RoboRace. On one hand, there's the HUGE advantage on developing AI for sim racing games, like you've mentioned, making our little niche even nicer than it is now. Also, whether you like it or not, driverless cars are a thing of the present and they're probably be mass implemented in the streets in no time, both for safety and convenience reasons. I believe that letting AI drive for drunk people or people with disabilities is a huge step forward.


However, there are issues with that. Coding is tremendously painful if not done right. I know that RoboRace has been years into development and they're probably not using Java to do it like broken college students, but what if they find limitations in the software or a bug in the code that would require rewritting the whole algorithm? That would sure be unpleasant and cost a whole lot. Not only that, but coding by itself takes time and resources.


Not only that, but they're also planning on pushing it (or, at least they were planning to, I haven't seen news on it for a while) driverless acceleration and skidpad testing for Formula Student competitions in Germany by 2021. That's a gigantic disadvantage to smaller budget teams, as most of them would not be able to get computer science students to do the job in that time frame. Not only that, but putting sensors all over the car and program them costs a lot. So FStudent teams would have three options:
>Invest a huge amount of sponsors' money into Driverless disciplines at the expense of being mediocre at the rest.
>Try to be in an advantage in driver events and doing the bare minimum to pass on driverless ones just so they aren't disqualified by not complying with the rules, basically transforming the engineering challenge and fun of building a self-driving car into a chore.
>Try to do well in both, therefore being in disadvantage at both and not good in either of them.


Of course, the paragraph above could be somewhat disregarded for RoboRace, but still, it makes me question some things: Is the focus of RoboRace to just build the fastest car or the best driverless AI? They are indeed better planned and better funded than usual FSAE teams, but how would it translate to the road? They've said that they used real racing drivers on DevBot to teach the AI the best racing lines around a track, but how would that translate to new tracks on their calendar? Sure, the AI won't learn them that way. Or even worse, how does that translate to the roads? Making people drive every road available to teach bots how to drive is, to say the least, unviable in so many ways, from mapping every road available to adding new ones, or ones that are blocked due to being under maintenance or being private areas. Lastly, will actual car manufacturers inject billions into cars that can drive themselves at the cost of low fuel efficiency, low durability parts, lower safety standards or simply huge prices that make those cars impossible for the regular market to buy, or will they simply cheap out on the electronics, making the drive a crappy, dangerous experience both for the driver and for pedestrians and other vehicles?


It has so much potential, but it also raises so many questions. Only time will tell, I think...

[SingleRacerSVR]

I'm a little confused here. Was I listening to the "usual Sim-Racing" Chris Haye....

Or was I listening to the silky smooth documentary voice... who yes, ALSO happens to be Chris Haye, Ha(ye), Ha(ye), Ha(ye ; )

[vMaxDaniel]

I think that an AI driver might have the potential to be faster if alone on track, ie. hot lapping. Sensors and servo motors have the potential to be a lot more precise, consistent and faster response than humans.

When racing wheel to wheel, I still believe that the human unpredictable behaviour will always take advantage to a AI program. At least until skynet is deployed and then we'll have to run for our lives

[sideswipebl]

No, WipEout is the future of racing. It has to be.