- Never drive on unfamiliar roads
- Never exceed the speed limit or follow too closely
- Slam on the brakes at the earliest sign of any trouble. If a human driver rear ends you in the process, it's their fault and will be reported as such.
- Never drive on the freeway --- i.e. at high speed where reaction times are more limited.
Bascially, only address a safe subset of what human drivers do and then delare victory.
> - Never drive on the freeway --- i.e. at high speed where reaction times are more limited.
The paper says human benchmark being used is only from the zip codes Waymo operate in, so they are quite comparable. In fact, they have this bit about including freeway mileage in the human benchmark:
> Waymo also almost exclusively operated on surface streets (non access-controlled freeways) with a unique distribution of driving that is representative of a ride-hailing fleet. In contrast, the benchmark represents the privately insured driver population that resides in these geographic regions. The associated benchmark mileage has more freeway driving than the Waymo ADS. There are several considerations when examining these results with respect to this limitation. First, freeway driving has a lower crash rate (Scanlon et al., 2024a). Including freeway driving makes this benchmark crash rate artificially lower, so, by including freeways in this study’s benchmark, the benchmark crash rate underestimates the true driving risk of where the Waymo ADS operates. Second, driving outside of these denser urban areas that the Waymo ADS operates would likely represent a reduction in overall relative crash risk.
> - Never exceed the speed limit or follow too closely
How is this a bad thing?
re: Speed limit -- Besides the obvious, it's because not every vehicle on the road is part of the lowest common denominator.
https://www.reddit.com/r/waymo/comments/1hfvb5o/waymo_visual...
>- Never exceed the speed limit or follow too closely
I would give credit to Waymo and Self-Driving systems for this. The promise of self-driving is that it won't do stupid things that humans may do that increase chance of accident.
Highways are so much easier. Everybody just wants to go straight. They do change lanes, but they usually know it's coming well before.
Hypothetically, if all the cars are synced, you eliminate certain congestion effects. E.g. traffic congestion travels backwards like a wave. But if all cars stop and start in unison, that's eliminated.
Likewise, synchronized cars can drive closer to each other, merge more efficiently, and reserve alternate routes.
Perhaps in the future I'll have enough trust in autonomous digital systems (and their designers) to close my eyes and walk across a street while high-speed traffic seamlessly weaves around me, but the sum of my experiences to date don't give me confidence that these sorts of systems wont occasionally fail in extremely catastrophic ways.
At least when I cross in front of a human driver I can ensure that they're looking me in the eye while they run me over.
I think you would be lucky to get such a courtesy. Most drivers probably wouldn't notice they ran you over unless it showed up in the stream they were watching on their phone while "driving".
I think you'll be surprised at how quickly you get use to self driving cars. They're technologically completely amazing, but about as exciting as an elevator in operation.
I don't know at what level you're joking, but this happens to humans all the time, for a variety of (compounding) reasons. Perceptual blindness.
And eventually its either illegal or exactly like in the 30-40, some car people will start to build rails, so you can't cross the street whenever you want.
Actually building safe infrastructure where bikes are save, is the actual solution. And we don't need fancy AI cars to it either.
Building technology that acts in the same space as biological beings obviously must acknowledge their indeterminate behaviour. Children, elderly, and pets exist and must be accounted for, a model of reality that assumes everyone following laws is flawed, all that. What you’re describing is obviously unrealistic and not what will happen.
Sure if you also remove, people, bikes and everything else. Congratulation, you just invented a grade separation. And if you are going threw all the effort of grade separation, you know what a good idea? To push as many people as possible along that route. And you know how you do that, by not putting each human into a huge 2t box of steel.
If you had the incredibly inefficiency of cars, to grade separate you would need to grade separate a HUGE space. And grade separation is incredibly disruptive. Grade separation is only worth if you can push lots of people threw.
> Likewise, synchronized cars can drive closer to each other
Because synchronized cars never have a mechanical failures or emergency break?
Here is an idea, you create these things called 'couplers' and then you physically couple multiple 'cars' together and then you actually can be really close together.
How about this, we move away from cars, and have beautiful parks for the people who live in these places. with nice restaurants having seats in what is now the road, with some nice not so fast trams going threw. And then some nice high density rail under or over on a guide-way.
Here is a good video going threw all these points and how great this scenario actually would be.
Moving from a 'Buffers and chain coupler' to advanced modern form of a 'Scharfenberg coupler':
https://www.youtube.com/watch?v=TQ7HlmPljyA
Hopefully in 10 years or so we don't have to listen to US train nerds bragging about the Janney couplers anymore.
To clarify, I'm talking about fluid-dynamic-like models of traffic congestion (oversimplification: traffic behaves like a compressible fluid). It's been a while since I studied this, but my impression is that synchronizing can eliminate some of these effects (perhaps make traffic more like a non-compressible fluid).
So I wasn't implying that people and bikes and mechanical failures would never stop traffic, but rather that traffic can better respond to those obstacles, and flow rate can be improved.
The realities of physics ensure that there will always be a delay and it will always lead to congestion as the pipe fills.
We route internet packets at around the speed of light and still experience traffic congestion on those networks.