GM Volt Forum banner

1 - 20 of 48 Posts

·
Administrator
Joined
·
20,191 Posts
After listening to a podcast on the Bolt self driving cars equipped with 40 sensors, 200MB/second data processing through multiple redundant systems, 3cm micro-mapping and labeling of cities and 40% of the Bolt parts being replaced with AV specific components, Id have to say that any talk about cheap self driving tech must be taken with a grain of salt.

Check out http://gm-volt.com/forum/showthread...-in-U.S.-cities-in-2019&p=4348793#post4348793
 

·
Moderator
Joined
·
8,680 Posts
The hard part of self driving is figuring out what the car should do any given moment, not making the car do it.

In terms of making the car do it, the second generation Volt is about the same as a modern Tesla (I think I read the new Volt has Bosch iBooster braking like Tesla, and I know they use a similar dual pinion electric power steering.)

In principle, that means that if someone came up with a box you could put on top of the car that had all the sensors and processors to figure out what to do well, it could be easily tied in to the car. I don't really expect to see such a box - certainly not before dedicated FSD cars appear.
 

·
Administrator
Joined
·
20,191 Posts
In principle, that means that if someone came up with a box you could put on top of the car that had all the sensors and processors to figure out what to do well, it could be easily tied in to the car. I don't really expect to see such a box - certainly not before dedicated FSD cars appear.

And today, that sensor and computer jammed box would likely be in the $100,000+ range.

Getting the price down requires mass production of the sensors, lower cost sensors and lot's of high-end data mapping (and more). Cruise Automation is working on that and has something like 1200 employees. They just bought a solid state LIDAR manufacturer and expect to drop LIDAR cost by 90%. They are testing in dense confusing city environments. Roll-out of fully self driving Bolt's is slated for 2019. To think that there is an easy, inexpensive bolt-on alternative is unrealistic.
 

·
Registered
Joined
·
1,521 Posts
And today, that sensor and computer jammed box would likely be in the $100,000+ range.

Getting the price down requires mass production of the sensors, lower cost sensors and lot's of high-end data mapping (and more). Cruise Automation is working on that and has something like 1200 employees. They just bought a solid state LIDAR manufacturer and expect to drop LIDAR cost by 90%. They are testing in dense confusing city environments. Roll-out of fully self driving Bolt's is slated for 2019. To think that there is an easy, inexpensive bolt-on alternative is unrealistic.
At least one that a company would want to be liable for if there was an issue.
 

·
Registered
Joined
·
3,717 Posts
I think the video is BS. It looks like a mostly straight road and the guy is steering with his left knee. Considering the limited success of autopilot and such with a whole bunch sensors, I suspect the whole openpilot thing might be BS.
 

·
Registered
Joined
·
3,717 Posts
After listening to a podcast on the Bolt self driving cars equipped with 40 sensors, 200MB/second data processing through multiple redundant systems, 3cm micro-mapping and labeling of cities and 40% of the Bolt parts being replaced with AV specific components, Id have to say that any talk about cheap self driving tech must be taken with a grain of salt.
Check out http://gm-volt.com/forum/showthread.php?306657

I'd say tons of salt. Or better, soap and water to clear away all the BS.
 

·
Registered
Joined
·
1,429 Posts
I think the video is BS. It looks like a mostly straight road and the guy is steering with his left knee. Considering the limited success of autopilot and such with a whole bunch sensors, I suspect the whole openpilot thing might be BS.
At best all he demonstrated was lane keep which the Volt already has. We are years away from having true self driving cars, what you are going to see in the next few years are cars that can drive themselves along interstates but off of limited access roads they will struggle until cars can communicate with each other in the same way that people do. Think about what happens at intersections, your actions are governed more my local custom and by looking at what other cars are doing then they are by the law. When the light turns green do you wait for the two cars that are going to run the red light to pass through before you start to move, or does everybody in line put their foot on the accelerator the second the light changes? It all depends on local custom. At a four way stop who gets priority? The law is specific about that but you don't know if the other people at the intersection know the law so you look at what they are doing before you start moving, if they are proceeding into the intersection then you wait, if they seem to be waiting for you, you proceed cautiously and then speed up if you see that they are respecting your decision. If two networked cars were at the intersection it would be easy for them because they not only would be programmed with the correct rules, they could signal each other, but if a lone intelligent car has to deal with several human driven cars it's a much harder problem. With the accidents that the prototype self driving cars have had you can see this illustrated. The self driving car is never technically at fault, it's always been the human driver. But if you think about it at a deeper level they are at fault because they don't behave like human driven cars, they strictly follow the law and they are cautious which confuses people because they expect that other drivers will at very least stretch the law.
 

·
Premium Member
Joined
·
14,156 Posts
I have a friend who started doing this with his Volt but a back ordered part and life has delayed it. I think it's crazy. People think a camera is like a human eye but it's really quite different.

In principle, that means that if someone came up with a box you could put on top of the car that had all the sensors and processors to figure out what to do well, it could be easily tied in to the car. I don't really expect to see such a box - certainly not before dedicated FSD cars appear.
If you listen to the podcast Steverino cited you'll find that the founder of Cruise Automotive wanted to do exactly this before he concluded it was too difficult. Plus the car wouldn't have all the redundant systems you want in an autonomous vehicle.
 

·
Moderator
Joined
·
8,680 Posts
I have a friend who started doing this with his Volt but a back ordered part and life has delayed it. I think it's crazy. People think a camera is like a human eye but it's really quite different.

If you listen to the podcast Steverino cited you'll find that the founder of Cruise Automotive wanted to do exactly this before he concluded it was too difficult. Plus the car wouldn't have all the redundant systems you want in an autonomous vehicle.
People talk a lot about needing redundancy for autonomous cars and it confuses me because it seems like a double standard - we're not requiring the same level of redundancy in all of the human driven cars on the road. I need redundancy where I can get it in aerospace, because I can't pull over to the side of the road (and that's a standard line folks use.) I think all I need for an autonomous car is predictable safe failure modes and hopefully the ability to get out of traffic.
 

·
Registered
Joined
·
605 Posts
I don't find it that unlikely since the car does have the hardware needed. But no matter how you look at it, your phone is driving your car. No thanks! I'll wait until the manufacturers integrate it. I can imagine that thing running down a half dozen pedestrians, cut scene, and 20 minutes later you are pointing at the phone hanging off your windshield with "OpenPilot has closed" on the screen and you pointing and saying "See... not my fault. My phone crashed".

Mike
 

·
Premium Member
Joined
·
14,156 Posts
People talk a lot about needing redundancy for autonomous cars and it confuses me because it seems like a double standard - we're not requiring the same level of redundancy in all of the human driven cars on the road. I need redundancy where I can get it in aerospace, because I can't pull over to the side of the road (and that's a standard line folks use.) I think all I need for an autonomous car is predictable safe failure modes and hopefully the ability to get out of traffic.
Well, if the ABS brakes fail, the human driver acts as the back up system and pushes the pedal, hard, engaging the mechanical brakes. Ditto for the steering. If you replace the human driver you need to find a substitute for the redundancy supplied by the driver.
 

·
Premium Member
Joined
·
14,156 Posts
We are years away from having true self driving cars
If you're talking about full autonomy everywhere then you are right. However, change the parameters and we either have full autonomous vehicles today or will have them in the near future. GM plans to roll out full autonomous vehicles in San Francisco and New York in Q1 of 2019. At a more limited scale, Waymo and GM have demoed full autonomous vehicles.

Now you won't be able to buy on of these vehicles. With the necessary lidar, radar, and cameras, these cars are going into ride sharing and fleets where the costs can be more easily absorbed.
 

·
Registered
Joined
·
3,440 Posts
Even if this system were just for lane keeping (I haven't researched it, but it seems to do at least that in the video), I think this could be a great thing to increase safety and reduce driver workload and fatigue. Especially when paired with ACC. We don't need full autonomy immediately in every car. The driver would still be fully responsible for seeing and reacting to every hazard, and obviously full handling in intersections and most other complex environments. He could do a better job of those high level tasks if not concentrating on lane keeping and spacing with the car ahead on freeways, especially on longer drives.

I know I would rather share the freeway with cars on this system vs. what I see human drivers doing every day, such as drifting across lane markers at the worst moment due to texting, etc.
 

·
Registered
Joined
·
4,011 Posts
People talk a lot about needing redundancy for autonomous cars and it confuses me because it seems like a double standard - we're not requiring the same level of redundancy in all of the human driven cars on the road. I need redundancy where I can get it in aerospace, because I can't pull over to the side of the road (and that's a standard line folks use.) I think all I need for an autonomous car is predictable safe failure modes and hopefully the ability to get out of traffic.
The difference is who gets held responsible. If you drive into a bridge abutment, it's your fault. If you build a self-driving attachment and it rolls your car into a bridge abutment, it's your fault. If you buy a gizmo and it sideswipes a bridge abutment, it's the MANUFACTURER'S fault. And they're going to do their damned best to make sure that doesn't happen through computer failure. And that means not having a failure situation which means having enough redundancy to make sure no failure leads to injury or loss.
 

·
Registered
Joined
·
1,056 Posts
You guys want self driving cars,,Gm can't even make a radio not freeze up,,good luck when one of these self driving cars fails and the driver is not paying attention.Hell,people around here don't pay attention now and they are in control.
will be a scary new world.
 

·
Super Moderator
Joined
·
6,277 Posts
I haven't tried it yet myself, but:
Video demo: https://www.youtube.com/watch?v=IYa5vQQEX2w
The port: https://github.com/commaai/openpilot/pull/104
The extra board needed: http://openboards.io/product/voltboard-vt/
There is a #volt channel on Comma.ai's Slack server: https://comma.slack.com/
I think people are VERY confused on the current purpose of OpenPilot. It is for Level 2 and equivalent to Tesla AP1 for highway driving (roadtrips) and stop and go traffic (commutes). This is HUGE actually for a lot of people and goes long ways for the percentage of time behind the wheel (especially boring time). Driving to the grocery store or out to dinner is a separate thing (Level 4/5)

OpenPilot is very cautious and even shuts off when you press the accelerator (unlike others including Tesla AP).

I don't find it that unlikely since the car does have the hardware needed. But no matter how you look at it, your phone is driving your car. No thanks! I'll wait until the manufacturers integrate it. I can imagine that thing running down a half dozen pedestrians, cut scene, and 20 minutes later you are pointing at the phone hanging off your windshield with "OpenPilot has closed" on the screen and you pointing and saying "See... not my fault. My phone crashed". Mike
and
I think the video is BS. It looks like a mostly straight road and the guy is steering with his left knee. Considering the limited success of autopilot and such with a whole bunch sensors, I suspect the whole openpilot thing might be BS.
Wait a second ... please understand they are using a particular type of phone, particular case (heatsink, fan), it has a dedicated OS on it, and they have their own hardware for the OBDII interface. It is for highway driving and stop and go as I mention below.

Note that CommaAI (OpenPilot) is putting the OS together to match the hardware (phone and OBDII) BUT the self-driving (Level2) software is MIT opensource for liability reasons.

---

If my wife's 2016 GEN II would have came with ACC I my have gone down this (OpenPilot) path at least for the fun of it. Of course, we know that GM could NOT even figure out how to get ACC on the early 2016s GEN IIand had to wait months later. (And as we know GM bought the self-driving expertise (Cruise) and couldn't even do it themselves.)

 

·
Super Moderator
Joined
·
6,277 Posts
I have a friend who started doing this with his Volt but a back ordered part and life has delayed it. I think it's crazy. People think a camera is like a human eye but it's really quite different.

If you listen to the podcast Steverino cited you'll find that the founder of Cruise Automotive wanted to do exactly this before he concluded it was too difficult. Plus the car wouldn't have all the redundant systems you want in an autonomous vehicle.
Be aware that just because the founder of Cruise failed does not prove anything. Keep in mind WHO this guy is and what his background is in regards to this OpenPilot project. See the first sentence below. Now he is focused on auto AI and supporting like 5 or 6 of the top 10 selling cars in the USA as a reasonable target. I don't know if he will succeed but it is fascinating to me!

http://reason.com/reasontv/2017/10/27/george-hotz-self-driving-autonomous-car

>>>
George Hotz, known online as GeoHot, became one of the world's most famous hackers at 17 when he was the first person to break into the iPhone and reconfigure it to be compatible with providers other than AT&T. He was also the first to jailbreak the PlayStation 3, allowing users to play with unauthorized software.

Now this 28-year-old technical wunderkind is up against Waymo, Tesla, Uber, and most of the auto industry in the race to build the first fully operational autonomous vehicle.

"I want to win self-driving cars," Hotz told Reason. Whereas Tesla and Waymo are developing complex systems with expensive LIDAR and other sensors, his company, Comma.ai, is trying to bring plug-and-play driverless technology to the masses. "We're running it on a phone," says Hotz.

He's taking an approach drastically different than his well-financed competition, and is operating with $3.1 million in seed money. Comma's dozen-member team, which works out of a residential house in San Francisco, has built technology that takes over the existing RADAR and drive-by-wire systems in modern cars, incorporates a smartphone's camera and processor, and then makes the car drive itself.

Hotz has a history of taking on tech titans, with mixed reactions. After the iPhone jailbreak, Apple co-founder Steve Wozniak sent him a letter of congratulations. After he hacked the PS3, Sony sued him. Hotz quickly became a cause celebre of so-called hacktivist groups including Anonymous and LulzSec. They attacked Sony's network, despite Hotz's protests, igniting a firestorm of legal and media scrutiny.

Comma.ai is Hotz's attempt to take on the big players in a new way. The company makes an app called Chiffr that turns a user's phone into a dashcam and monitors its GPS and accelerometers. Now Comma is launching Panda, an open source, $88 dongle that plugs into the car, links it to the phone, and puts out fine-grain detail about every aspect of a drive. Hotz ingests all the data from Chiffr and Panda users and feeds it to his artificial intelligence system, which then learns how to drive.

According to Hotz, this approach gives him significant advantages over competitors such as Waymo. His network is entirely crowdsourced and running on some of the most popular cars in the country. He doesn't need to build another expensive, specially designed vehicle and employ a trained driver and an engineer every time he wants to add another data point. And all his data come from real-world experience.

Hotz says Waymo and others take a rule-based approach to driving that doesn't reflect the reality of how people operate cars. "The humans ain't changing to match the self-driving spec," he says. "In order to really get access to the full, diverse spectrum of what driving is, you need a huge crowdsourced database."

While Tesla's training model is more closely aligned with his, Hotz says the company will similarly be restricted to the high-end market. He got into a public spat with Tesla founder Elon Musk in 2015, after Hotz says the mogul changed the terms of a deal for him to build a better vision system for Tesla's Autopilot than the one supplied by partner company Mobileye. Musk claims Hotz bragged that he could build a better system, and then welched on the bet.

"All I said was I could build a better vision system than Mobileye, myself, in 3 months," replies Hotz. "And I kind of did that."

Despite his new gig, however, Hotz maintains a hacker's spirit. He has a bounty program that pays out $10,000 to customers who are able to port Comma's software and enable it to tap into their car's driving systems. Hotz, however, denies that this constitutes hacking, even though he offers the bounty instead of contacting carmakers to access their APIs.

<snip>

<<<
 

·
Registered
Joined
·
3,717 Posts
Well, if the ABS brakes fail, the human driver acts as the back up system and pushes the pedal, hard, engaging the mechanical brakes. Ditto for the steering. If you replace the human driver you need to find a substitute for the redundancy supplied by the driver.
If ABS fails you better remember how to maintain directional control. Just pushing the pedal harder isn't the answer. Not sure what you mean about the steering. If the power assist fails, you can use more muscle. If the steering itself fails the best you can do is hope to stop safely.
 
1 - 20 of 48 Posts
Top