Autonomous Car Progress

[diplomat33]

@mspisars I just downloaded the Waymo app and set it up and it says I am ready to go as soon as I am in the service area.

 

[mikes_fsd]

In fact, anybody can download the Waymo One app now. If you are not in the geofenced area, you will get a message saying you need to be in the area. But if you download the app today and go to the service area Phoenix, you can summon a driverless ride any time.

They did change this, I checked back during Thanksgiving break and it was still telling me I had to be approved.
Today it says that I can use it in the "Metro Phoenix" area.

 

[Snuffysasa]

Since November the waitlist was removed and anyone in the area can use, even visitors.

 

[diplomat33]

They did change this, I checked back during Thanksgiving break and it was still telling me I had to be approved.
Today it says that I can use it in the "Metro Phoenix" area.

Correct. I just downloaded the app. It says the same thing. It says that I am good to go when I am in "the metro phoenix area".

 

[DanCar]

I'm doubtful that you can get a ride without a safety driver easily. There are reports of people using the service and getting a safety driver. The number of videos without a safety driver are surprisingly tiny. Hopefully that will change soon.

 

[mikes_fsd]

It says that I am good to go when I am in "the metro phoenix area".

That's a step in the right direction.

I'm doubtful that you can get a ride without a safety driver easily. There are reports of people using the service and getting a safety driver. The number of videos without a safety driver are surprisingly tiny. Hopefully that will change soon.

That is the ~$3.5 billion dollar question ( I think Google spent more than that, but this is the number I found online Companies Have Spent $16 Billion on Self-Driving-Car Research )

Note that same article states:

Waymo, the self-driving moonshot from Alphabet (a.k.a. Google's parent company), has spent about $3.5 billion testing self-driving Chrysler Pacificas in Silicon Valley and Phoenix. It has a "self-driving" ride-hailing service. No one is at the wheel, but it's monitored remotely, and there's a chase vehicle in case the van gets confused and a human has to leave the chase car and get in the van to take over.

I did not know that info in the bolded portion... So for every driverless "robo-taxi" they have a chase vehicle... this seems silly! I hope it is not the case now.

 

[diplomat33]

I did not know that info in the bolded portion... So for every driverless "robo-taxi" they have a chase vehicle... this seems silly! I hope it is not the case now.

We've seen the chase vehicle for some rides. I don't know if it is for every single robotaxi.

For consumer cars, it is not an issue since the driver can take over. But in the case of driverless robotaxis, you need a plan if the car does get stuck. You can't just leave the customer stuck in a robotaxi on the side of the road. I am sure Waymo will wean off using chase vehicles over time as the ride-hailing gets better.

 

[johnmodely]

We've seen the chase vehicle for some rides. I don't know if it is for every single robotaxi.

For consumer cars, it is not an issue since the driver can take over. But in the case of driverless robotaxis, you need a plan if the car does get stuck. You can't just leave the customer stuck in a robotaxi on the side of the road. I am sure Waymo will wean off using chase vehicles over time as the ride-hailing gets better.

Im sure they can just remotely take over in that case...needing someone to chase and manually take over sounds silly when the thing has cameras and an internet connection.

 

[diplomat33]

Im sure they can just remotely take over in that case...needing someone to chase and manually take over sounds silly when the thing has cameras and an internet connection.

The chase cars are not for taking over driving. And Waymo actually does not remotely take over. Waymo uses remote observers that can give the car a hint if it gets stuck but the car stays in autonomous mode. We've seen that in videos. The chase cars are more like Tesla mobile service. They can get involved if there is some type of physical problem with a driverless car where remote guidance would not work.

 

[MP3Mike]

Im sure they can just remotely take over in that case...needing someone to chase and manually take over sounds silly when the thing has cameras and an internet connection.

We have seen cases where the rider had to wait for the "roadside assistance van" to arrive, and a driver from it got in and "un-stuck" the Waymo. (Waymo customer service stayed on the line with the rider until the driver arrived.)

 

[Snuffysasa]

I'm doubtful that you can get a ride without a safety driver easily. There are reports of people using the service and getting a safety driver. The number of videos without a safety driver are surprisingly tiny. Hopefully that will change soon.

Not true! You can get a ride without a safety driver easily. Every public ride is driverless and they are readily available! There was just a day or 2 during rain where they added safety drivers temporarily.

But yes the number of videos on the internet is small. That is just because the driverless rides are boring, and people are not traveling / going out due to covid.

 

[diplomat33]

But yes the number of videos on the internet is small. That is just because the driverless rides are boring, and people are not traveling / going out due to covid.

There are a few youtubers that are riding in Waymo for the thrill or to share content about autonomous driving. But most of the riders are regular folks actually using the Waymo ride to get somewhere they need to go. So they don't care about recording their rides. I think that explains the lower number of videos. Although, the videos have increased a lot since Waymo lifted the NDA and opened it up more.

It also occurs to me that there is a sort of paradox maybe with autonomous driving. We are excited now about robotaxis because they are uncommon and new. But when autonomous driving gets good enough and common enough, it will become boring. I mean, we don't get excited about getting into a regular taxi. Eventually, we will get so used to robotaxis just like riding in a normal taxi.

 

[daniel]

Isn't it about time to say something about the true cause of those speed limits? If they are, in fact, too low, why is there no political pressure to raise them?

In Germany in densely populated areas the reason is sometimes noise reduction for the people living next to the street. So we cannot just discuss drivability.

These are highways where noise is not an issue. There's no political pressure to raise the speed limits because the limits are not being enforced. So they don't affect drivers of cars other than Teslas. If Tesla would allow autosteer to operate at 10 mph over the limit, I'd still be driving slower than most, but it would be less obnoxious. So instead I just don't use autosteer when the traffic is very heavy.

... [a] CEO who makes crazy promises that he can't keep.

 

[diplomat33]

This thread might suit your needs better:
Autonomous Car Progress
@diplomat33 will help you with your arguments.

There is a fine difference I am not sure if I made it clear. I really am not so familiar with Lidar, only know that it has draw backs, such as it cannot penetrate fog, and will treat plastic bag as solid material.

But can anybody say that they know 100% of FSD in Tesla and that Lidar will add absolute no value to it down the road?

My point is not so much about whether Lidar is good or bad. My point is if Tesla decide to add Lidar later on, then it is a show of strength, not so much of an engineering failure.

AFAIK, lidar can have some difficulty in rain and fog depending on the density of the rain/fog and the wavelength of the lidar.

But lidar has some big advantages over cameras, making it a really good sensor to have for both driver assist and full autonomy:

"“Lidar technology is inherently superior to camera and radar in certain performance aspects that are crucial for avoiding forward collisions and which support a move within the industry to implementing lidar as a crucial sensor for ADAS applications.”"

"Lidar performs free-space detection more efficiently and precisely than cameras, by providing real-time measurements of how far surrounding objects are from the vehicle, with no additional computational processes or sensors required. As a result, data from a single lidar sensor directly provides the fundamental building block of a successful driver assistance system: accurate free-space detection. That is, lidar utilizes precise distance measurements of surrounding objects to map areas where it is safe for the vehicle to drive.

In contrast with lidar, camera-centric approaches require multiple sensors and complex computational processes to infer distance of surrounding objects and thereby determine safe driving paths."

Revolutionizing driver-assistance systems with forward-looking lidar, part II

This is why many automakers are now including a forward facing solid state lidar located in the front bumper. They are cheap, are hidden in the front bumper and add reliability in detecting hazards or objects in front of the car.

Ultimately, all sensors will have pros and cons. There is no single sensor that works perfectly all the time, in all conditions. This is why on page 47, the "Safety for Automated Driving" document recommends using multiple sensor types:

"As of today, a single sensor is not capable of simultaneously providing reliable and precise detection, classifications, measurements, and robustness to adverse conditions. Therefore, a multimodal approach is required to cover the detectability of relevant entities. In more detail, a combination of the following technologies shall provide suitable coverage for the given specific product:
CAMERA - Sensor with the highest extractable information content as sensor captures visible cues similar to human perception. Main sensor for object/feature type classification. Limited precision in range determination, high sensitivity to weather conditions.
LIDAR - High-precision measurement of structured and unstructured elements. Medium sensitivity to environment conditions.
RADAR - High-precision detection and measurement of moving objects with appropriate reflectivity in radar operation range, high robustness against weather conditions.
ULTRASONIC - Well-established near-field sensor capable of detecting closest distances to reflecting entities.
MICROPHONES - Public traffic uses acoustic signals to prevent crashes and regulate traffic, e.g. on railway intersections. Thus, devices capturing acoustic signals are required for automation levels where the systems need to react to these."

So, I agree that if Tesla added lidar in the future, it would not be a sign of failure. It would simply mean that they recognized advantages in lidar to compliment their existing sensor suite.

 

[diplomat33]

IMO, this blog from "Let's Talk Autonomous Driving" has a good definition for "fully autonomous driving":

"Fully autonomous driving technology is designed so that a human driver does not take over the maneuverability of a vehicle. You do not need a license to take a ride in a vehicle with autonomous driving technology, and your hands never need to touch a steering wheel."

I know we've debated at length the different SAE levels. But perhaps, we can agree on this definition for "fully autonomous driving"?

The blog also has some good stats:

- An Insurance Institute for Highway Safety (IIHS) study conducted in 2019, half of respondents believed a driver-assist feature allowed them to drive hands-free, even though these systems require drivers to keep their hands on the steering wheel at all times.

- A more recent study from AAA Foundation found similar results. When drivers thought they were behind the wheel of a “semi-autonomous'' vehicle, more than half of them spent 30% more time with their hands off the wheel than people who did not think they were in a semi-autonomous vehicle.

According to these stats, it appears that a lot of drivers are engaging in potentially risky driving habits because they think that "driver assist" can do more than it actually can.

Source: We've changed our name. Here's why. | Let's Talk Autonomous Driving

Edit: I will try to use that definition from now on. It is the best definition I've seen and it seems consistent with how the SAE and other AV companies define the term.

 

[Mardak]

"Fully autonomous driving technology is designed so that a human driver does not take over the maneuverability of a vehicle. You do not need a license to take a ride in a vehicle with autonomous driving technology, and your hands never need to touch a steering wheel."

Seems like plenty of vehicles fall into this definition for years/decades, e.g., airport people movers. Do there need to be additional descriptors/qualifiers to separate out what Waymo is doing vs the old vehicles vs those that are more capable than Waymo?

 

[diplomat33]

Seems like plenty of vehicles fall into this definition for years/decades, e.g., airport people movers. Do there need to be additional descriptors/qualifiers to separate out what Waymo is doing vs the old vehicles vs those that are more capable than Waymo?

Well, the definition aims to clarify the differences between driver assist and autonomous driving in order to prevent "autonowashing" (when driver assist systems are mistaken for autonomous driving). The definition is very generic so yes, you would probably need to add extra info about the ODD to give a more complete description of a specific AV product.

I think the definition does a good job of describing what autonomous driving is, in general terms. But yes, I would agree that you would need to add extra descriptors/qualifiers in order to provide a more complete description of an AV product. And you would need more descriptors/qualifiers if you wanted to compare two different AV products. Obviously, if two products are both "fully autonomous" under this definition, you would need something else to distinguish them.

 

[diplomat33]

CES 2021 will have a session on "Self-Driving Vehicles Moving Forward: Who Will Set the Rules?" on Wednesday, Jan 13, 1:40pm to 2:20 EST.

Description:
With the coronavirus pandemic increasing demand for contactless operation and delivery, self-driving vehicles provide a safe mode of transportation. Governments around the world are moving forward to set the rules for driverless cars as they become more of a reality than a myth. Hear innovators and policy makers discuss what consumer expectations and government decisions mean for the future of mobility and national competitiveness.

Home

I don't know if the session will be streamed live or recorded. Hopefully, there will be something after the session for those who cannot attend CES.

 

[Snerruc]

I posted in another thread that I’m not as smart as you guys, but there are lots of things a human can do that I don’t see autonomous cars doing in the foreseeable future. People think out of the box to recover from problems. They can find houses without an address or not on a map easily. Coming up on a blocked road they can back up to the nearest intersection and reroute. Also they can create a road when it’s necessary. I could go on and on. People are amazing creatures.

 

[diplomat33]

Waymo has announced it will stop using the term "self driving" because it's become misleading, like "autopilot". Instead it will say "fully autonomous" to make it clear that it is talking about cars which completely drive themselves without human oversight or intervention.

Yep. I am glad Waymo is doing this.

Here is the blog in question:
Waypoint - The official Waymo blog: Why you'll hear us saying fully autonomous driving tech from now on

And here is the blog from Let's Talk Autonomous Driving:
We've changed our name. Here's why. | Let's Talk Autonomous Driving