Welcome to Tesla Motors Club
Discuss Tesla's Model S, Model 3, Model X, Model Y, Cybertruck, Roadster and More.
Register

Waymo

This site may earn commission on affiliate links.
Interesting waymo interaction with other drivers on a narrow road, pulling over to the side to make way for oncoming traffic.


It shows Waymo is actually considering how much space the other vehicle needs to pass safely, not just it's own, and can pull over to give the other vehicle the space it needs. That's advanced behavior. And it does it very smoothly too.
 
It shows Waymo is actually considering how much space the other vehicle needs to pass safely, not just it's own, and can pull over to give the other vehicle the space it needs. That's advanced behavior. And it does it very smoothly too.
Some of those streets are ridiculously narrow with double parked cars on both sides of the streets. I've been spoiled by where i live, its mostly wide streets.
 
  • Like
Reactions: diplomat33
Some of those streets are ridiculously narrow with double parked cars on both sides of the streets. I've been spoiled by where i live, its mostly wide streets.

Yes, some of those streets are very narrow. And I think narrow streets are a good example of why HD maps and lidar are beneficial. The fact is that sometimes, you need that cm level precision because the space is just so tight. Certainly it is helpful if you want confident, smooth driving. Without that precision, the AV is more likely to hit something, or at the very least, be very hesitant, jerky, or get "stuck".
 
  • Funny
Reactions: bkp_duke
Yes, some of those streets are very narrow. And I think narrow streets are a good example of why HD maps and lidar are beneficial. The fact is that sometimes, you need that cm level precision because the space is just so tight. Certainly it is helpful if you want confident, smooth driving. Without that precision, the AV is more likely to hit something, or at the very least, be very hesitant, jerky, or get "stuck".
I don't see how HD maps help with this. You need to precisely know your distance relative to the un-mapped parked and moving cars. Your precise location on the road doesn't really matter. LIDAR is best for this. Ultrasonic sensors are second best.
 
I don't see how HD maps help with this. You need to precisely know your distance relative to the un-mapped parked and moving cars. Your precise location on the road doesn't really matter. LIDAR is best for this. Ultrasonic sensors are second best.

I agree lidar is best for the precise distance measurements. HD maps are probably not super essential for this type of scenario. The reason I mention HD maps is because of localization. If you can precisely locate yourself on a cm precise map, it could help the car maneuver since you will know exactly where you are relative to say the curb or a side street etc... And with parked cars blocking the view, your cameras and lidar may not see the curb. so having that knowledge on a precise HD map could help IMO.
 
I agree lidar is best for the precise distance measurements. HD maps are probably not super essential for this type of scenario. The reason I mention HD maps is because of localization. If you can precisely locate yourself on a cm precise map, it could help the car maneuver since you will know exactly where you are relative to say the curb or a side street etc... And with parked cars blocking the view, your cameras and lidar may not see the curb. so having that knowledge on a precise HD map could help IMO.
This is correct. It's not just a matter of seeing and measuring the distance to the other car but contextualizing the other road agents and their action so you can navigate around them. I'm not saying you cannot do that with just live sensors but a prior is valuable.

Localization provides the 3D pose of a self-driving car inside a high-definition (HD) map, including 3D position, 3D orientation and their uncertainties. Unlike the use of a navigation map with GPS, which only requires an accuracy on a scale of a few meters, the localization of a self-driving car has a much higher requirement of accuracy relative to the map, usually on the order of centimeters and a few tenths of degrees.

The value of localizing to a HD map might not be immediately obvious, given the rapid progress in live perception. However, when localized to a HD map, the car can make decisions based on information beyond its current field of view.

Localizing to a HD map also allows a self-driving car to contextualize the behaviour of other nearby vehicles within the map. At an intersection for example, it can recognize that an oncoming car is in a left turn lane and will not proceed straight, making it safe to perform an unprotected left turn

 
  • Like
Reactions: diplomat33
Axios spent a whole day with Waymo in Phoenix:

Highlights:

I spent a whole day testing Waymo's capabilities. I asked its cars to drop me off on a one-way downtown street, pick me up at the airport, go to the eastern edge of its boundaries, and navigate construction projects.
  • Within about five minutes, my nerves dissipated. I was enthralled by the ghostly sight of a steering wheel turning on its own, and impressed as the car maneuvered around obstacles.
The big picture: Rides were typically $3-$5 less than a comparable Uber trip — and there was no driver to tip.
  • In the morning, I never waited more than seven minutes for a ride. But when I requested a vehicle from downtown Mesa around 5 p.m., I had to wait 25 minutes.
  • The airport drop-off and pickup were a breeze. It took me 12 minutes to get from the car to Terminal 4 using the Sky Train
Yes, but: Waymo doesn't use freeways (yet), which turned what should have been one 30-minute ride into a 55-minute haul.
  • The vehicle is trained to find a clear, safe place to pull over — in one instance, about a five-minute walk from where I was trying to go in downtown Phoenix.
  • And the expanded service boundaries are still limiting. Waymo covers only about half of downtown Mesa, which made for a less-than-fun quarter-mile walk in 97-degree heat.
1 bizarre thing: My first ride from Chandler to downtown Phoenix took a wildly ineffective, zigzagging route through several industrial parks and neighborhoods.
  • I figured the cars were programmed to avoid busy streets, but those I took later in the day had no problem with the likes of Jefferson Street and Mill Avenue.
The bottom line: Waymo isn't quite convenient enough to replace Uber yet — but it's easy to see how it could get there soon.

Source: https://www.axios.com/2023/06/21/waymo-driverless-car
 
Last edited:
My take: I think it is a very good sign that Waymo is typically cheaper than Uber. And the fact that Waymo is doing thousands of rides per day, including all kinds of city driving, with no incident is a good indicator that safety is very good. So the tech part is solid. I think the big challenges now are mostly operational: Waymo needs to reduce wait times and trip times in order to increase convenience. To do that, Waymo needs to add highways, add more vehicles and expand the geofence even further. IMO, when they do those things, Waymo should have a very solid and competitive ride-hailing service that they can scale to other cities.
 
What does the car do if there is no room to pull off to the side? Say there is a drop off at the edge of the road

The car would search for a better spot. The car is programmed to look for the best possible spot to pull over, hopefully as close as possible to the destination. if no good spot is available close to the destination, the car will search for another spot. In some cases, this might mean driving a bit further to find a good spot.
 
According to this article, Waymo is in in early discussions to expand service to SF airport but no official announcement yet:

"Autonomous driving tech company Waymo is planning a southern expansion of its driverless taxi service to a well-known and high-traffic destination: the San Francisco International Airport. A lobbyist for the company, which is headquartered in Mountain View, first began having conversations with SFO officials regarding the matter in March, public records show. SFO spokesperson Doug Yakel confirmed Thursday that airport stakeholders are in discussions with Waymo, but said that "no decisions have been made." He said that there is not yet a timeline for the service expansion to the airport "at this early stage."

Source: https://www.bizjournals.com/sanfrancisco/news/2023/06/23/waymo-robotaxi-expansion-s-f-airport.html

So nothing official but I think expanding to SFO would make sense. Waymo is already servicing the Phoenix airport. Waymo has said that providing rides to airports is part of their roadmap. I think for SF, the main obstacle will be regulatory since SF officials are opposed to Waymo.

Also, the fastest way to get to SFO is via highways so if Waymo does add service to SFO, it it is very likely that Waymo will also add highways.
 
  • Like
Reactions: spacecoin