I think you're over-playing how decisively a Waymo will move and under-playing how decent the average human is.
I've ridden in Waymos. They don't exactly slap on the blinker and move at the limit of traction like someone about to miss their exit. If cut off they absolutely will go full brake rather than perform any sort of spicy lane change or turn.
> If cut off they absolutely will go full brake rather than perform any sort of spicy lane change or turn.
Essentially, a meat driver was waiting at a stop sign to make a turn onto the main road. I was in a Waymo driving on the main road and did not have a stop sign.
When we were 10 meters away from the intersection, the meat driver suddenly started to enter the intersection. I have no idea why.
Full brake would've hit the other car in the driver's side door at 40 km/h.
> under-playing how decent the average human is.
I got to SMFC in CSGO which means I'm in the top 3% of players in clicking on heads within 500 ms of them appearing on my screen. I have never reacted as fast as that Waymo did.
If instant (<50ms) reaction would have lowered the speed only to 40km/h in 10m, Waymo was going too fast for the intersection IMO.
My experience is that for a human driver to react quickly in city driving conditions, style and prep are more important than reaction time: in the case you describe (entering an intersection with another car waiting on a stop sign perpendicular to your path), I'd have my foot hanging over the brake and off the gas pedal — this has helped me avoid hitting many other cars with inattentive/distracted/bad drivers, and even pedestrians running over the road or a red light on a crosswalk. When you are prepared and looking, you slam the brakes much faster!
>Waymo was going too fast for the intersection IMO.
This is literally impossible without slowing to single digit speeds for every intersections. At some point you just have to rely on the other traffic honoring signaling and signage or having some desire for self preservation.
Waymo uses new cars which probably have 100km/h (62mph) to zero stopping distance around 36m (120ft) — that's what my 2020 car quotes and tests at. As stopping distance grows quadratically, from 50km/h it would have stopped in 8m. Two lane street is usually at least 8m wide.
The claim was that after braking for 10m, it was still going at 40km/h. It'd take another 6-7m to come to a full stop. If it was a full 18m stopping distance (half the one from 100km/h), that'd mean a bit over 70km/h, so over 60km/h anyway for 16-17m.
I do not know of any country where there are intersections you can go through at 60+ km/h legally.
This does not mean that Waymo in question was going too fast, but something is off in the claim (maybe it did not react on time and really brake for 10m; maybe the collision speed was not a full 40km/h; or maybe it was going too fast...).
>I do not know of any country where there are intersections you can go through at 60+ km/h legally.
The US is dotted with "real highways" (i.e. designed as such, not a former main street that's seen a bunch of upgrades) with 50+mph speed limits and low traffic streets that tee into them with nothing more than a stop. And this isn't some middle america thing that can be dismissed as backwards flyover states. The rich coastal states have them too. Divided medians and T-junction type are fairly common. 2-way stops and cross type junctions less so in my experience but in more rural areas they're more common.
>This does not mean that Waymo in question was going too fast, but something is off in the claim (maybe it did not react on time and really brake for 10m; maybe the collision speed was not a full 40km/h; or maybe it was going too fast...).
Seems like someone pulled out from a residential road onto a main road with no f's to give and the waymo went around and OP is messing up the numbers a little. No matter how fast you're going it always feels faster from the passenger seat.
It's a huge difference between someone merging into or crossing a high speed road.
In most of Europe at least (I did not drive enough in States to remember), non-highway intercity roads have a speed limit of around 50-55mph, but where there is any merge or crossing, this is reduced to 30-35mph.
To be honest, I don't know how to convert freedom units to km/h.
I based my estimate on the Waymo going slower than other cars and the city speed limit in Toronto being 50 km/h, and I took a stab at the numbers. I think it was Beverly Blvd which the internet says has a speed limit of 35 mph = 56 kph at the time?
I was in the back seat so I couldn't judge the distance that well. I passed the other car in a second so I guessed that's how much ground I covered.
Understood: I wasn't trying to nitpick the numbers, but simply showing that you were likely crediting Waymo a tad too much for what it has done.
Human drivers correct each others' mistakes every single day, and we don't hear about them because... well, nothing has happened. The argument could be made that Waymos will make fewer mistakes, so fewer evasive maneuvers will be needed, but it's great to hear that Waymo's performance is coming closer to bringing good human driver capability with faster reaction time enabled by tech.
I appreciate you converting to SI units, but I am ok with you keeping them as-is too: either side can do the conversion to whatever they need, and HN is very much a mixed audience.
Pulling out randomly, I see it all the time. I beat the computer by a) anticipating; and b) assuming other drivers are idiots who don't see me. I don't have to calculate trajectories and whatnot, people aren't computers, and they can do some things better than a computer can, especially a solely-reactive one.
To be fair, we are not provided with the sensors to swerve safely.
If we had some sort of 360 constant recording in the car (on screen?) it would be safer for humans to swerve. Instead we have to move our head, which is cheaper but lacks info.
That's why we now have rear cameras
We have rear cameras because people DONT move their head. And because regulations have made cars way taller than they need to be, meaning there is a big blind spot close to the ground
I mean, even in low cars you cannot see a small enough kid walking behind your car. That's why you back slowly.
Back when I just got my driver license, there is a big lesson many drivers go through (in Italy) which is you back off a parking and there is an obstacle that's so low that cannot be see through the back window and it's small enough that cannot be seen through the mirror. You hit it and if you followed the "go slow part" you only damaged the paint.
So I'm not opposing the ideas of rear cameras, but I'm totally against tall cars, because you cannot see kids IN FRONT either now.
Depends on how you define "average driver": what if 95% of the crashes are caused by 5% of the drivers?
My reading of all the human crash stats has been that majority of them happen when human drivers are impaired (drunk, drugged or too tired): as this is something we could (in theory, at least) control, I'd like to see and compare with stats for non-impaired human drivers too.
Then, I'd like to see it compared to attentive, non-distracted drivers too (but we won't have crash data for this, as they would avoid most potential crashes).
Note that I am only talking things under every human driver's control, and not things like skill, reaction time, etc.
Also, modern cars (like Waymos) will have a much lower braking distance compared to "average": eg. my Volvo has 35m braking distance from 100km/h or 62mph compared to 50m (45% more) listed as average (excluding reaction distance) — so from 50km/h, it should be around 8m!
It certainly matters: "average driver" does not exist, and 95% of the drivers beat the average.
So a claim how autonomous driving system beats the average would only tell us that it beats 5% of the human drivers.
Now, the way stats are massaged here is not even about "drivers", but miles driven, and this language is even worse. We'd need to make sure we are looking at human-driven miles in the same area, same roads, with similar cars.
>To be fair, if 5% of drivers cause 95% of crashes then the average driver is still terrible.
>The median one might be better, but does it even matter? The average driver is still wreaking havoc.
Yes it matters. To be acceptable this technology needs to be at least in the same ballpark as a median-ish person on a median-ish day. Not some nonexistent average that is pulled down by the 1/X people who are drunk and the 1/Y who are from Socal and driving in Maine in a blizzard.
The fact that you basically never hear of "average non criminal driver" or "median law abiding driver" and that there is no real attempt at even standardizing a concept of normal drivers not engaged in bad behavior just reeks.
It's like the door is intentionally being left open for the same slight of hand as when people peddle some policy goal having to do with school shootings and back it up with statistics that are mostly normal crime. Or they are peddling some devious tax that will screw a whole lot of people, and they justify it with an average that's dragged way up by a few oddballs, or dragged way down by a bunch of zeros. Seems like the safety crowd and and self-interested industry are setting up to play off each other in a "recyclable plastic" sort of way.
Second off, what are you talking about that the "average driver is wreaking havoc"? The average driver is filing a collision claim every 15-20yr depending on who's numbers you believe. While I don't know the distance between average and median, either is a fairly high bar that Waymo and friends have to meet.
But the point is if we get all the median and better drivers off the road and replace them with autonomous vehicles, yet keep the worst 5% on the roads too, we are potentially worse off.
In Houston the stats suggest that every driver should get into a crash at least once every. But many ppl haven’t been an crash all their lives and more have been in multiple
I've ridden in Waymos. They don't exactly slap on the blinker and move at the limit of traction like someone about to miss their exit. If cut off they absolutely will go full brake rather than perform any sort of spicy lane change or turn.