A total funding of $16.8 million sounds like a tiny amount of money to be able to develop this kind of complex technology, especially anything related to cars, which are heavily regulated. Their competitors - Tesla, Google and the huge car manufacturers - all have billions of dollars to throw around.
"The product to date has been a system that converts existing cars into self-driving cars down highways. ... There are sensor units that go on top of the car near the windshield and then there are actuators that control the steering and driving. There’s also a computer that goes in the trunk... The plan had been to sell Cruise systems for $10,000 each."
This doesn't sound like it would get significant acceptance from the public. Are people going to be comfortable getting third-party modifications done on their cars, which would be of unknown quality and safety and surely invalidate their cars' warranties?
I may be wrong, but I have a feeling that there is already a very large market of people who like to tinker with cars. Imagine a self-driving Delorean (https://en.wikipedia.org/wiki/DeLorean_DMC-12)
I agree, it doesn't sound like much money. Their poaching of Tesla's head of Autopilot R&D is very interesting though, and is definitely a sign of confidence.
I think you're making quite a few assumptions though: (a) Getting third party modifications to make your current car self-driving may look crazy today, but when self-driving cars become more of a thing, it'll certainly be a more attractive option when compared to buying a new car altogether. (b) Right now the quality and safety are unknown, but assuredly they won't be in the future. I'm sure these devices will be tested thoroughly and have to meet specific regulations. (c) I'm not sure they would void the warranty, it would depend on a lot of factors.
Regardless, the article hints at an upcoming pivot from Cruise, so the described product may not even hit the market.
That sort of company usually gets picked up by existing manufacturers. Like perhaps Chinese or Indian manufacturers who will be late to the game. It's a good idea to go this route.
All "Cruise" seems to be talking about is lane keeping plus radar cruise control. BMW, Buick, Cadillac, and Volvo had that years ago. Mercedes is shipping it now. The big problem is keeping drivers from relying too much on a dumb system that almost works[1].
Everyone with a clue in the industry realizes the problem. Semi-automatic driving, with the expectation that the driver is alert and ready to take over, is an accident waiting to happen. Mercedes has a device to check if hands are on the wheel. (See video below for someone defeating that.) Cadillac has a seat vibrator. Most of those semi-auto systems are unable to detect junk such as a mattress on the highway, deer, pedestrians, and work crews. They assume a benign freeway environment. Most won't engage in a non-freeway environment.
What bothers me about "Cruise" is that they promote this assist technology as a full self-driving system, like what Google and CMU/Cadillac have in test. See the Cruise ad.[2]
The NTSB defines four levels of self-driving:
0 - manual
1 - Function-specific Automation
(example: anti-collision radar braking.)
2 - Combined Function Automation
(adaptive cruise control plus lane centering.)
(current Mercedes, etc. level)
3 - Limited Self-Driving Automation
(still needs a driver available to take over.)
(current Google level)
4 - Full self-driving automation
(driver does nothing, may not even be present)
(Google is trying to get there for 25MPH cars.)
Cruise is at level 2, but advertises as if at level 3.
Driving a level 2 car and thinking it's a level 3 car is deadly.
We completely agree with your safety concerns (along with those of many others we've spoken to in the industry). That's one of the reasons we haven't released a product that works in the way currently advertised on our website.
While NHTSA has begun thinking about autonomous vehicle safety, there is still much to be done. We plan to release some of our thoughts about this in the next couple of months.
That seems to be exactly what Cruise is doing. Watch the video. "Seamlessly takes you wherever you want to go". "Just sit back and enjoy the ride". The website: "Just sit back and watch as the RP-1 guides your car down the road." Watch them drive on the Great Highway at SF's beach, which has traffic lights, pedestrian crossings, and bicyclists.
Suggested reading: Federal Trade Commission truth in advertising rules.[1]
Just like we learn how to use cruise control in certain driving situations, I would expect drivers can learn how to rely on "Supercruise" in appropriate situations.
Specifically I think the most beneficial use-case would be the mind numbing highway stop-and-go crawl which millions contend with daily. If my car could follow the car in front of me with lane-keeping, it seems like that would be extremely useful. IMO, it's not a system that needs to be able to accelerate past even 25 mph before reverting to a more typical / unassisted speed adaptive cruise control. But IMO it must have an extremely snappy almost aggressively close follow distance, since the computer should have better reflexes. In traffic, if it's leaving even a car length, it would drive me crazy.
I would assume these systems are challenged by a lack of intuition around traffic flow, and the [in]ability to steal glances down the line of cars (I guess depending on how high-mounted the top camera is). If the only real input is "mm to next bumper @ 60Hz" can you do better than a human driver?
I think another key trick is in really nailing the "My Car" transition cue. I guess it could be some sort of touch sensor on the wheel, or a flick of a paddle shift, but I think the best input might be a tap on the gas pedal.
Finally found a video of it in action [1]. During this 1:11 video the car dings an alert for "apply steering" no less than 3 times? The follow seems a bit sluggish but not as bad as I was expecting.
Found another example which shows the lame follow distance and sluggishness. [2] Hilariously, there's an accident at the end of the video in the next lane over due to the stop-and-go traffic. But in this form, I think these things are basically traffic creation devices.
"I may be wrong, but I have a feeling that there is already a very large market of people who like to tinker with cars. Imagine a self-driving Delorean"
Or my own version, a level 3-ish self-driving 1977 VW Rabbit :)
"Vogt couldn’t change lanes because of California state law around autonomous vehicles"
This statement is complete bullshit. The California DMV doesn't have any clear details defining operations of vehicles at this point in time. They only have rules about what is required to test on public roads. This includes 5 million dollar bond, applying for permit, etc...[1]
This statement alone (if correct from the reporter) makes me nervous about the amount of spin going on.
To clarify, the California DMV carefully defines an "Autonomous vehicle" but has carve outs for collision avoidance systems [1]. While adaptive cruise control and lane keep assist are mentioned explicitly as collision avoidance systems, automatic lane changing is not. So we weren't comfortable adding that feature until the DMV published their regulations for the operation of Autonomous Vehicles (which still hasn't happened) and we met those requirements.
Exactly how have they taken the lead here? They bought a bunch of bolt-on stuff and slapped together some software that works in a very limited way. Am I missing something?
Can someone explain how Google's long running car program fits in? Does Googe license patents (or map data, or...) to companies? Is Google car a PR thing that isn't competitive with the auto industry? Is Google hoping to leapfrog over all the "assisted driving" tech to be the first fully autonomous car?
22 comments
[ 3.1 ms ] story [ 59.4 ms ] thread"The product to date has been a system that converts existing cars into self-driving cars down highways. ... There are sensor units that go on top of the car near the windshield and then there are actuators that control the steering and driving. There’s also a computer that goes in the trunk... The plan had been to sell Cruise systems for $10,000 each."
This doesn't sound like it would get significant acceptance from the public. Are people going to be comfortable getting third-party modifications done on their cars, which would be of unknown quality and safety and surely invalidate their cars' warranties?
I think you're making quite a few assumptions though: (a) Getting third party modifications to make your current car self-driving may look crazy today, but when self-driving cars become more of a thing, it'll certainly be a more attractive option when compared to buying a new car altogether. (b) Right now the quality and safety are unknown, but assuredly they won't be in the future. I'm sure these devices will be tested thoroughly and have to meet specific regulations. (c) I'm not sure they would void the warranty, it would depend on a lot of factors.
Regardless, the article hints at an upcoming pivot from Cruise, so the described product may not even hit the market.
Everyone with a clue in the industry realizes the problem. Semi-automatic driving, with the expectation that the driver is alert and ready to take over, is an accident waiting to happen. Mercedes has a device to check if hands are on the wheel. (See video below for someone defeating that.) Cadillac has a seat vibrator. Most of those semi-auto systems are unable to detect junk such as a mattress on the highway, deer, pedestrians, and work crews. They assume a benign freeway environment. Most won't engage in a non-freeway environment.
What bothers me about "Cruise" is that they promote this assist technology as a full self-driving system, like what Google and CMU/Cadillac have in test. See the Cruise ad.[2]
The NTSB defines four levels of self-driving:
Cruise is at level 2, but advertises as if at level 3.Driving a level 2 car and thinking it's a level 3 car is deadly.
[1] https://www.youtube.com/watch?v=Kv9JYqhFV-M [2] http://www.getcruise.com/
While NHTSA has begun thinking about autonomous vehicle safety, there is still much to be done. We plan to release some of our thoughts about this in the next couple of months.
Wait, what? That sounds like (awkward) double-speak. So you're advertising a product you don't/won't have?
Suggested reading: Federal Trade Commission truth in advertising rules.[1]
[1] https://www.ftc.gov/news-events/media-resources/truth-advert...
Specifically I think the most beneficial use-case would be the mind numbing highway stop-and-go crawl which millions contend with daily. If my car could follow the car in front of me with lane-keeping, it seems like that would be extremely useful. IMO, it's not a system that needs to be able to accelerate past even 25 mph before reverting to a more typical / unassisted speed adaptive cruise control. But IMO it must have an extremely snappy almost aggressively close follow distance, since the computer should have better reflexes. In traffic, if it's leaving even a car length, it would drive me crazy.
I would assume these systems are challenged by a lack of intuition around traffic flow, and the [in]ability to steal glances down the line of cars (I guess depending on how high-mounted the top camera is). If the only real input is "mm to next bumper @ 60Hz" can you do better than a human driver?
I think another key trick is in really nailing the "My Car" transition cue. I guess it could be some sort of touch sensor on the wheel, or a flick of a paddle shift, but I think the best input might be a tap on the gas pedal.
Found another example which shows the lame follow distance and sluggishness. [2] Hilariously, there's an accident at the end of the video in the next lane over due to the stop-and-go traffic. But in this form, I think these things are basically traffic creation devices.
[1] - https://www.youtube.com/watch?v=r9RAn0DwAAU [2] - https://www.youtube.com/watch?v=TmdWBH-IvDg
Or my own version, a level 3-ish self-driving 1977 VW Rabbit :)
https://www.youtube.com/watch?v=iCE35zbkciU
This statement is complete bullshit. The California DMV doesn't have any clear details defining operations of vehicles at this point in time. They only have rules about what is required to test on public roads. This includes 5 million dollar bond, applying for permit, etc...[1]
This statement alone (if correct from the reporter) makes me nervous about the amount of spin going on.
[1] : http://www.dmv.ca.gov/portal/wcm/connect/d48f347b-8815-458e-...
[1] : https://www.dmv.ca.gov/portal/dmv/detail/vr/autonomous/bkgd
Not revolutionary, but can be very useful nevertheless.