In your analogy the driver would have stopped and literally had no idea what to do until the human passenger informed them---and it's likely not "hey turn right on ash street as it's a shortcut in rush hour" type of driving, but "driver stops and says uh, there is construction and a policeman and signs and complicated lane lines, help! where do I go" and the passenger had to specifically tell show them which lane to go in when.
-
Want to remove ads? Register an account and login to see fewer ads, and become a Supporting Member to remove almost all ads.
Waymo
- Thread starter Daniel in SD
- Start date
diplomat33
Average guy who loves autonomous vehicles
Oddly enough, there are weeks in my neighbourhood when this happens hundreds of times per day. The person behind the wheel is completely incapable (or at very least unwilling) to continue without assistance from a 3rd party giving them directions. The city hires ~50 people to give directions during this period (street party for ~1 million, roughly 25 streets are closed to traffic).
The person in control of the vehicle, in the drivers seat, is still legally the driver despite requiring assistance.
Last edited:
diplomat33
Average guy who loves autonomous vehicles
Yes and some human drivers have to do that too, especially new drivers who are still learning and don't have a lot of driving experience. Don't let the hype fool you, yes, AVs are getting better and are much better than they used to be just a couple years ago but the fact is that even our best autonomous driving, is still basically like a new human driver that is still learning, even though AVs are gaining more experience every day. Waymo has the most autonomous driving experience so they are able to handle city driving pretty well but it cannot handle every single case. So it still needs some remote assistance.
Last edited:
powertoold
Active Member
Was in SF today, Waymos are ubiquitous throughout the city, almost all of them driverless and driving in touristy areas. It looks great. It also looks very expensive to maintain, as I also passed by a manned garage for parked Waymos. I assume the cars also get cleaned inside and outside on a routine or daily basis.
On the other hand, I didn't see a single Cruise car today. Good riddance.
The Waymos look so good, I was motivated to download the app and join the wait-list lol. It's great that Waymo is paving the way for true driverless taxis.
On the other hand, I didn't see a single Cruise car today. Good riddance.
The Waymos look so good, I was motivated to download the app and join the wait-list lol. It's great that Waymo is paving the way for true driverless taxis.
Drive by wire is a term for electronic systems that either augment or replace traditional mechanical controls in a vehicle. Instead of using cables, linkages or hydraulic pressure to actuate vehicle controls, you use electronic actuators and motors to activate, steering, brakes and fuel/engine. You are still providing the input necessary to turn the wheels, apply brakes, accelerate but there is no longer a hydraulic or mechanical linkages between the user control input and the wheels, brakes, etc. It has nothing to do with the level of automation, there is very little connection between the two except for the fact that it involves a vehicle.
J3016 addresses all these things. There is no need for all these pedantry and false equivalences if you just take the time to read the actual document.
Remote Assist : At L4, the ADS is still responsible for performing the DDT and OEDR. There is no such thing as L3.5
Example Zoox has 2 kinds of remote assist.
Path Suggestion: The remote assistance operator drops a suggested paths for the ADS but the ADS is responsible for driving the vehicle using the suggested path and taking other road users into consideration which means following appropriate lanes, traffic rules, monitoring the surrounding, applying the steering and braking. At no point does the remote assist control the vehicle. The L4 ADS is always engaged and in control.
Collaborative Guidance: The remote assist operator gives extra context to the scene, like for example labeling a stopped vehicle as a blockage if the ADS is confused whether that particular vehicle is indefinitely stopped at a traffic light in a turning lane or just waiting for the traffic light to change to make a turn.
Remote Driving is not Remote Assist and having remote assist does not change the level of driving automation. A L3 ADS can have a remote driver do the fallback because the remote driver is performing the DDT and OEDR. A L4 ADS when engaged does the entire DDT and OEDR regardless of it needs remote assistance to decide the optimum path out of a situation.
Remote driving
How about we stop talking about scenarios that are not likely to happen. There are always people monitoring these systems to ensure they are operating safely. It is also a fact that if the tire fell off the vehicle and there was no one to change the tire the vehicle would be stopped.
J3016 addresses all these things. There is no need for all these pedantry and false equivalences if you just take the time to read the actual document.
Remote Assist : At L4, the ADS is still responsible for performing the DDT and OEDR. There is no such thing as L3.5
Example Zoox has 2 kinds of remote assist.
Path Suggestion: The remote assistance operator drops a suggested paths for the ADS but the ADS is responsible for driving the vehicle using the suggested path and taking other road users into consideration which means following appropriate lanes, traffic rules, monitoring the surrounding, applying the steering and braking. At no point does the remote assist control the vehicle. The L4 ADS is always engaged and in control.
Collaborative Guidance: The remote assist operator gives extra context to the scene, like for example labeling a stopped vehicle as a blockage if the ADS is confused whether that particular vehicle is indefinitely stopped at a traffic light in a turning lane or just waiting for the traffic light to change to make a turn.
Remote Driving is not Remote Assist and having remote assist does not change the level of driving automation. A L3 ADS can have a remote driver do the fallback because the remote driver is performing the DDT and OEDR. A L4 ADS when engaged does the entire DDT and OEDR regardless of it needs remote assistance to decide the optimum path out of a situation.
Remote driving
Can you provide some quote from Waymo that supports that statement? Waymo doesn't have people monitoring the steering of the car. They monitor the fleet and provide high level instructions like "safe to go around" to the car if stuck (which happens very seldom).
diplomat33
Average guy who loves autonomous vehicles
0:00 Pick up
0:26 Start of ride
0:39 Taking right of way from privileged cross traffic
1:44 No reaction to irrelevant active fire truck
6:24 Slowing for unknown
8:45 Good tracking of light rail
10:27 Passing vehicle parked in bike lane
11:22 Unprotected left turn with oncoming
11:38 Hi Waymo
11:57 Slowing for work zone maybe
12:33 Nudging into oncoming lane with traffic
14:00 Jerky while passing large vehicle
14:19 Entering alley
15:14 Pull over
I was on a 3 lane road this morning. It was all empty as everyone was at church.
The road has lot of repair going on. The section that is repaired is fresh concrete, almost grey white and a big contrast to the dark color of the age old road.
As I approached one of those new sections, there were no cars around me, including none ahead or behind me. FSDb suddenly slowed down and came to a stop. I then had to tap on the accelerator to keep going.
It is interesting that the change in the color of the road made it stop. On days when there is traffic, it does not stop when it is following another vehicle. But alone, it stopped. Interesting behavior.
EVNow
Well-Known Member
Only time it's empty here is during Superbowl
I've posted before how difficult it is to understand the lanes when the road is wet and the Sun is low and shining. You see all the patches on the road and none of the lane markings.
diplomat33
Average guy who loves autonomous vehicles
Driverless Waymo in busy Chinatown. Lots of interesting scenarios:
0:00 Pick up
0:12 Start of ride
0:48 No reaction to irrelevant siren
0:59 Merge in intersection
1:44 Yielding to pedestrian entering street
3:26 Nudging for large vehicle
4:34 Brake hopping at red light
8:15 Pulling forward to stop line
9:56 Entering Grant Ave in Chinatown
10:01 Nudging for person in street
10:40 Yielding to several pedestrians in crosswalk
11:05 Entering narrow alley
11:53 Unprotected turn with pedestrians crossing
12:49 Yielding to person in street
13:20 Tourists: up close and personal
14:08 Hard brake for unknown (bad pedestrian prediction?)
14:47 Several narrow gaps
16:09 Passerby asks if Waymo can be overridden
16:40 Right turn into cursed intersection design
18:07 Yielding to parallel parking vehicle
19:13 Waiting for intersection to clear
20:25 Lane change with traffic
21:18 Stopping for fire station keep clear zone
22:16 Slowing for aggressive cut-in
22:49 Abrupt braking at stop sign
23:41 Waiting for intersection to clear
24:00 Nudging for pedestrian entering street
26:17 Passing stalled vehicle
26:24 Pull over
0:00 Pick up
0:12 Start of ride
0:48 No reaction to irrelevant siren
0:59 Merge in intersection
1:44 Yielding to pedestrian entering street
3:26 Nudging for large vehicle
4:34 Brake hopping at red light
8:15 Pulling forward to stop line
9:56 Entering Grant Ave in Chinatown
10:01 Nudging for person in street
10:40 Yielding to several pedestrians in crosswalk
11:05 Entering narrow alley
11:53 Unprotected turn with pedestrians crossing
12:49 Yielding to person in street
13:20 Tourists: up close and personal
14:08 Hard brake for unknown (bad pedestrian prediction?)
14:47 Several narrow gaps
16:09 Passerby asks if Waymo can be overridden
16:40 Right turn into cursed intersection design
18:07 Yielding to parallel parking vehicle
19:13 Waiting for intersection to clear
20:25 Lane change with traffic
21:18 Stopping for fire station keep clear zone
22:16 Slowing for aggressive cut-in
22:49 Abrupt braking at stop sign
23:41 Waiting for intersection to clear
24:00 Nudging for pedestrian entering street
26:17 Passing stalled vehicle
26:24 Pull over
3 rounds of layoffs in a year can't be great for morale at Waymo.
sfstandard.com
Tech Layoffs: Google's Waymo Robotaxi Firm Slashes More Jobs
Waymo confirmed the company's third round of layoffs this year.
sfstandard.com
diplomat33
Average guy who loves autonomous vehicles
3 rounds of layoffs in a year can't be great for morale at Waymo.
Tech Layoffs: Google's Waymo Robotaxi Firm Slashes More Jobs
Waymo confirmed the company's third round of layoffs this year.
No, it can't be great. I assume the new President of Alphabet is responsible for the layoffs because she is probably ordering Waymo to cut costs. Frankly, I am a bit surprised that Waymo still has two co-CEOs. That seems top heavy to me. I would think that with the focus on cost cutting, that they would go back to just one CEO position. Could we see Dolgov and/or Mawakana get fired, to be replaced with a single CEO that is more business oriented? Could Dolgov go back to being CTO since tech is his strength and Mawakana stay as CEO? Personally, I feel like Dolgov has done better than Mawakana because Waymo seems to have made more progress on his side of things (improving tech) than on her side of things (business scaling). But perhaps, Dolgov and Mawakana have enough clout, as top execs, that both their jobs are safe.
It will be interesting to see how things unfold. Waymo clearly has great AV tech. The big question is what will happen from a business point of view. Can Waymo leverage the tech into a profitable business? One option might be for Waymo to cut back on adding new cities, cut some jobs, and focus on consolidating their current ride-hailing in Phoenix, SF and LA into a profitable service. Personally, I would love to see Waymo look to licensing their tech. Waymo could sell their custom hardware to carmakers looking for high quality sensors. Ans surely, Waymo could leverage their software for quality driver assist or semi-autonomous driving features. For example, they could use a piece of the Waymo Driver to offer reliable lane keeping, collision avoidance, auto lane change, etc... I imagine carmakers looking for reliable driver assist could be interested in Waymo software. So I think that would be a great way to monetize their tech. I just think Waymo would be wise to look for side revenue to help pay for the scaling of robotaxis. That is why I think the Mobileye strategy is so smart because they are able to generate nice revenue with their ADAS tech to help pay for their development and deployment of AVs. I love Waymo tech but to be honest, if their strategy is "lose billions until we hopefully make a profit someday when we have scaled", that's not a business strategy that is going to fly at the Alphabet Board meeting. I think a few years back, when the tech was more primitive, Waymo could kind of get away with the argument of needing to spend/lose billions to develop the tech. But now that the tech works well in 3 cities, that argument does not work anymore. It's time to start showing a real business model.
Last edited:
diplomat33
Average guy who loves autonomous vehicles
Well, I've seen numerous tweets from Waymo riders saying they prefer Waymo over Uber/Lyft because Waymo offers the better rider experience and that prices are competitive with Uber and Lyft. So there is some anecdotal evidence at least that Waymo is viable or close to viable now against Uber & Lyft. But I think Waymo needs to do a lot more scaling in terms of number of vehicles in order to be economically competitive with Uber & Lyft. But I also wonder how much does Waymo really need to beat Uber & Lyft to be viable? I don't think Waymo needs to supplant every single Uber and Lyft vehicle. If Waymo can achieve some profit per ride that will be a great start. And if Waymo, which offers a better rider experience than Uber and Lyft, can also offer affordable rides, even on a more scale, I would say that Waymo would be viable. Remember, viable just means that Waymo can stay alive against Uber and Lyft. Viable is not the same as beating Uber and Lyft.
The cost of a standard Uber vehicle is very low. The vehicle can be as as old as 20 years old or brand new. It doesn't matter. The average would be 20K. Not so for Waymo. That is where it gets challenging.
I'm really hoping Waymo doesn't end up like a lot of Google's other "moonshot" ideas.
IMO Google Glass might be a good corollary to Waymo. A very developed product that, at least to me, was on the verge of being if not revolutionary. But killed by public perception (if Cruise has anything to say about it) and a lack of "want" to continue the project (currently not convinced on this one, but the x3 layoffs this year make me wonder).
Similar threads
- Article
