'Human Vs. Autonomous Car' Race Ends Before It Begins (arstechnica.com) 26
A demonstration "race" between a (human) F1 race car driver Daniil Kvyat and an autonomous vehicle was just staged by the Abu Dhabi Autonomous Racing League.
Describing the league and the "man vs. machine" showdown, Ars Technica writes, "Say goodbye to the human driver and hello to 95 kilograms of computers and a whole suite of sensors." But again, racing is hard, and replacing humans doesn't change that. The people who run and participate in A2RL are aware of this, and while many organizations have made it a sport of overselling AI, A2RL is up-front about the limitations of the current state of the technology. One example of the technology's current shortcomings: The vehicles can't swerve back and forth to warm up the tires. Giovanni Pau, Team Principal of TII Racing, stated during a press briefing regarding the AI system built for racing, "We don't have human intuition. So basically, that is one of the main challenges to drive this type of car. It's impossible today to do a correct grip estimation. A thing my friend Daniil (Kvyat) can do in a nanosecond...."
Technology Innovation Institute (TII) develops the hardware and software stack for all the vehicles. Hardware-wise, the eight teams receive the same technology. When it comes to software, the teams need to build out their own system on TII's software stack to get the vehicles to navigate the tracks. In April, four teams raced on the track in Abu Dhabi. As we've noted before, how the vehicles navigate the tracks and world around them isn't actually AI. It's programmed responses to an environment; these vehicles are not learning on their own. Frankly, most of what is called "AI" in the real world is also not AI.
Vehicles driven by the systems still need years of research to come close to the effectiveness of a human beyond the wheel. Kvyat has been working with A2RL since the beginning. In that time, the former F1 driver has been helping engineers understand how to bring the vehicle closer to their limit. The speed continues to increase as the development progresses. Initially, the vehicles were three to five minutes slower than Kvyat around a lap; now, they are about eight seconds behind. That's a lifetime in a real human-to-human race, but an impressive amount of development for vehicles with 90 kg of computer hardware crammed into the cockpit of a super formula car. Currently, the vehicles are capable of recreating 90-95 percent of the speed of a human driver, according to Pau. Those capabilities are reduced when a human driver is also on the track, particularly for safety reasons....
The "race" was to be held ahead of the season finale of the Super Formula season... The A2RL vehicle took off approximately 22 seconds ahead of Kvyat, but the race ended before the practice lap was completed. Cameras missed the event, but the A2RL car lost traction and ended up tail-first into a wall...
Khurram Hassan, commercial director of A2RL, told Ars that the cold tires on the cold track caused a loss of traction.
Describing the league and the "man vs. machine" showdown, Ars Technica writes, "Say goodbye to the human driver and hello to 95 kilograms of computers and a whole suite of sensors." But again, racing is hard, and replacing humans doesn't change that. The people who run and participate in A2RL are aware of this, and while many organizations have made it a sport of overselling AI, A2RL is up-front about the limitations of the current state of the technology. One example of the technology's current shortcomings: The vehicles can't swerve back and forth to warm up the tires. Giovanni Pau, Team Principal of TII Racing, stated during a press briefing regarding the AI system built for racing, "We don't have human intuition. So basically, that is one of the main challenges to drive this type of car. It's impossible today to do a correct grip estimation. A thing my friend Daniil (Kvyat) can do in a nanosecond...."
Technology Innovation Institute (TII) develops the hardware and software stack for all the vehicles. Hardware-wise, the eight teams receive the same technology. When it comes to software, the teams need to build out their own system on TII's software stack to get the vehicles to navigate the tracks. In April, four teams raced on the track in Abu Dhabi. As we've noted before, how the vehicles navigate the tracks and world around them isn't actually AI. It's programmed responses to an environment; these vehicles are not learning on their own. Frankly, most of what is called "AI" in the real world is also not AI.
Vehicles driven by the systems still need years of research to come close to the effectiveness of a human beyond the wheel. Kvyat has been working with A2RL since the beginning. In that time, the former F1 driver has been helping engineers understand how to bring the vehicle closer to their limit. The speed continues to increase as the development progresses. Initially, the vehicles were three to five minutes slower than Kvyat around a lap; now, they are about eight seconds behind. That's a lifetime in a real human-to-human race, but an impressive amount of development for vehicles with 90 kg of computer hardware crammed into the cockpit of a super formula car. Currently, the vehicles are capable of recreating 90-95 percent of the speed of a human driver, according to Pau. Those capabilities are reduced when a human driver is also on the track, particularly for safety reasons....
The "race" was to be held ahead of the season finale of the Super Formula season... The A2RL vehicle took off approximately 22 seconds ahead of Kvyat, but the race ended before the practice lap was completed. Cameras missed the event, but the A2RL car lost traction and ended up tail-first into a wall...
Khurram Hassan, commercial director of A2RL, told Ars that the cold tires on the cold track caused a loss of traction.
I guess Max (Score:4, Funny)
Verstappen doesn't have to worry about being replaced by AI.
Dunno about Checko Perez though
Re: (Score:2)
Verstappen doesn't have to worry about being replaced by AI.
Dunno about Checko Perez though
It would improve post race interviews though.
living organisms can adapt (Score:5, Interesting)
The algorithm is easy, recognizing the exceptions makes software design, difficult. All intelligence suffers from "you don't know what you don't know" but thinking organisms can adapt depending on experience, thinking machines can't.
Re: (Score:3, Interesting)
but thinking organisms can adapt depending on experience, thinking machines can't.
Not yet they can't. There's no reason a thinking machine can't adapt to changing situations because they already do. Think of airliners and how the software already makes adjustments to flight characteristics without human intervention. Yes, those characteristics are pre-programmed, but the system still has to react to the changing situation.
I'm certain there are machines in manufacturing which make adjustments on the fly ba
Re:living organisms can adapt (Score:5, Informative)
I remember watching a demonstration/report of people using a generalist self-driving AI and having it learn how to drive on ice/snow. It very quickly figured it out.
Adaptive algorithms are well known, covering things like having a drone figure out what power level is needed to keep to a steady altitude, even when commands to increase power to the motors demonstrate a delay before lift changes.
Re:living organisms can adapt (Score:4, Insightful)
Didn't the Gran Turismo Devs already demonstrate all of this? Tyre warming, adaptation, an AI that players can't beat?
It's the sensor suite that gives them the necessary data that is the issue.
Re: (Score:2, Informative)
Re: (Score:2)
The algorithm is easy, recognizing the exceptions makes software design, difficult. All intelligence suffers from "you don't know what you don't know" but thinking organisms can adapt depending on experience, thinking machines can't.
This is why I'm not worried about AI drivers taking over any time soon. We don't have proper AI (Artificial General Inteligence) we've just got a computer that can respond to a large set of pre-programmed scenarios... So when an unexpected issue occurs it can't respond to it, most of the time it doesn't even know something is wrong until after it happens. The big problem with computers is that they'll make the same mistake over and over again until someone fixes it, they're incapable of learning.
As someo
I mean it's a sport (Score:5, Insightful)
I mean I'm pretty sure that a mechanical crane could beat any weightlifter you care to name...
Re: (Score:2)
Re: (Score:3)
It won't be much fun watching a 500 lap race when you know the finish is going to be within 0.00001 seconds.
Obligatory Futurama [youtu.be].
Re: (Score:2)
Re: (Score:1)
a mechanical crane could beat any weightlifter you care to name...
Clark Kent.
Re: (Score:2)
Clark Kent.
You're going to need St. Anselm's proof of existence on that one ...
Re: I mean it's a sport (Score:2)
Only because Senna is not on this world any longer.
Interesting. (Score:1)
AI works by dividing the entire problem space into cells, one cell corresponding to one action.
If you cannot put things like detecting traction issues and compensating into such a grid, and the implication is that you can't, then AI cannot handle this problem.
AI will never replace drivers (Score:5, Insightful)
AI will never replace drivers in Formula 1 for the simple reason that no one would want to watch it. They had traction control and banned it because it was making the racing boring and predictable, and just a stupid competition of who could spend more money. AI drivers would be no different.
Although a lot of people would watch the first race, and a significant number of people would watch multiple races, this would fall off very, very quickly. No one really wants to see a parade of perfect laps by every car. I'm sure there's a few people on here who would watch the AI races every week, but you are in a tiny, tiny, tiny minorit, no where near the 450 million viewers they have now.
People don't even want to see AI competitions in computer games, Twitch is dominated by humans playing video games, even though bots can play them faster and more accurately. We want to see the mistakes, we want emotion, and we want interesting things happen in the race.
When Charles LeClerc won the Italian Grand Prix in a Ferrari this year, thousands of fans (probably over 10,000) filled the main straight for the trophy award. There would be zero reason to ever do this again with AI, no one i going to cheer the programmer. You do not kill that kind of enthusiasm for your own sport.
F1's recent surge in popularity is mainly due to the Netflix series "Drive to Survive" which focuses on the drivers and their relationships with their teams and other drivers. This entire surge would be lost immediately if they replaced the drivers with AI, not to mention the fans would already watched F1 for the same reason before Netflix got involved.
F1 is not a racing organization trying to build and race the fastest cars in the world, it is an entertainment organization that uses building and racing the fastest cars in the world to get more fans at the track, viewers at home, and sponsorship money. They are not going to throw those away to showcase AI for exactly zero benefit to them, since they are not an AI company, it would be financially suicidal.
Re: (Score:2)
If AI racing makes sense at all, what makes sense is to do the AI race days immediately after the human race days. You can do it on days when it's not convenient for humans for whom it's not their job, and people can watch any interesting parts later. Racing only "makes sense" for other than entertainment value when it's helping R&D, and it's not clear that this is actually valuable to any automakers as it's not very applicable to what their vehicles have to do regularly.
Re: (Score:2)
do the AI race days immediately after the human race days
I might be pretty lazy. But I can still get up and change the channel.
Re: (Score:2)
I might be pretty lazy. But I can still get up and change the channel.
Some people will want to see it, but not enough people to justify having racing series just for them. Having the computer cars' activity tied in with the humans' means keeping costs down. The track is already prepared, and all of the people who need to be on site are already there, the cameras are in place, etc etc.
Re: (Score:2)
Unfortunately, this would throw the insane schedule and logistics of F1 into chaos. They run 31 races across 5 continents, it would be a massive undertaking to keep people and equipment an extra day. It would cost less but would be still millions of dollars wasted per race event. When they finish a race on Sunday, they frequently have to tear everything down and move it to another country, then set it back up again (it is so, so, much more than the cars and garages), and be ready to go by Thursday afternoon
Would have been useful to consult a driver (Score:1)
> Khurram Hassan, commercial director of A2RL, told Ars that the cold tires on the cold track caused a loss of traction.
It's a poor designer who blames his tools.
Unexplained assertions (Score:2)
One example of the technology's current shortcomings: The vehicles can't swerve back and forth to warm up the tires.
What? Why not?
It's impossible today to do a correct grip estimation.
What? Why not?
Seriously, there's nothing inherently magic about swerving a car to warm up tires or gathering sensor data to determine grip. So, the car didn't have that today, but why couldn't it have it tomorrow?
Three levels of failure (Score:2)
1. Their testing plan didn't lead up to this by, for example, have Autonomous cars race Autonomous cars to learn the track, successful passing maneuouvers and zones, and demonstrate them.
2. That would have allowed machine learning to lower their lap times to near human performance.
3. They used Danil Kvyat. Next time use a real F1 driver.
Stupidity all around, but maybe the UAE will arrest more visitors who tap an airport employee on the arm, or UK vacationers who kiss on the beach. You can sugarcoat human