Making eye contact with drivers makes riding a lot easier. You know they see you.
(But - twice in 20 years I’ve had the chilling experience of a driver making eye contact and pulling out in front of me anyway.)
That was more of a problem in my 'Sturm & Drang' time. These days grannys on their electically-assisted bicycles are faster than me, hence I don't surprise car drivers any more.
It would be nice if there were a simple directional light on the top of robotaxis that indicates where it’s intending to go next (directional light) and how imminently it will begin moving (perhaps how fast the light blinks) when e.g. at a stop sign.
I was very harsh about the extremely poor interactions with Cruise, but I have yet to see anything to sour me with Waymo.
They have started trialing exactly this as of about two weeks ago. [edit: But I haven’t seen it operate up close yet.]
I don’t need to check if they see me. They do.
With such complexity and unknowns within the codebase - there is likely going to be a much higher amount of variability w.r.t. safety and AI decision making. I'm not so sure I'd categorically say a self driving car from say Nikola or some SPAC'd startup would make me feel all that safe.
(Weird this was down voted, just sharing personal experiences)
I live right near Oracle Park where there was a big Cruise Yard, and there is a ton of traffic and cruise, zoox, etc self driving cars circling the neighborhood endlessly.
It's not as if I've had a overtly terrible single incident. It's as though they're causing unsafe driving conditions for pedestrians, cyclists and other motorists regularly.
Driving unreasonably slow and erratic, stopping in the middle of streets, blocking traffic, not responding to the need for them to move or respond as human driver would.
They're problematic. I could start a youtube channel or something it happens enough in my day to day life.
I bike by Oracle park almost every day.
Until you're being dragged under them.
Maybe they should take a note from the Apple Vision Pro and display a pair of eyes on the windshield looking at you? (/s ?) :)
https://www.theverge.com/2023/10/13/23913251/waymo-roof-dome...
Million dollar idea right there
What do you mean by that? Do you think if their cars hurt someone or cause damage to property you can't sue them? What is the basis of this belief?
They're still trying to do this weird universal signage when they need some way to replicate eye contact and 1-to-1 signaling. I feel safe biking or driving around other cars because I can directly communicate and negotiate with other traffic participants. I can't do that with an automated car as they're currently designed, and I feel like that's a huge flaw.
Waymo cars are extremely, extremely aware of their surroundings. They have a category, distance, speed, and distribution of possible future paths for all cars, pedestrians, and bicyclists within sight over 360 degrees.
> Why don't they have a (or something like a) 360 degree string of LEDs (for example on the roof) that lights up a few in the direction of human obstacles if they're close enough?
I don't think it's so obvious this helps. What is the threshold for how far away something is before it's indicated on the ring? If I'm standing just outside that distance, how will I know it's because of distance rather than because I haven't been seen? If the distance is set long, how will I know it sees me and not something behind me?
The natural way to get around these problems is for the car to "point out" everything it see using, e.g., fake eyes. But now you have a car covered in eyes surveilling the neighborhood, and people don't like that either.
I feel like this is a non-issue. The LEDs can be a certain size. They could be dimmable so their brightness can be modulated by distance.
All the things you mention are an issue with real drivers too, but we make it work because we can signal and communicate. Give these cars the same functionality and experiment. This isn't an insurmountable issue, but it is one that needs to be addressed that isn't being addressed at all. Throwing up your hands and saying it's too difficult is less than helpful. They're putting these things on the road with people. They have a responsibility to figure this stuff out. Especially if they want to go beyond a handful of cities.
No one is throwing up their hands. Human-autonomous-vehicle communication is a whole research field. Waymo has every incentive here to both make the vehicles safer and to make pedestrians feel safer (holding actual safety fixed).
The ultimate test is not your first gut reaction as to whether they have done "enough", but rather it's whether the vehicles are safer than human-driven ones, and the evidence so far strongly points to that being the case.
Often you can't tell if a human sees you. Even in day, they may have tinted windows, or the sun might make it impossible. At night, you definitely can't tell.
If it's someone at a 4-way stop, you can usually tell when they come to a stop far from back the crosswalk. Otherwise I don't walk out in front of them. I also try to be aware if I'm in their blind spot.
https://dl.acm.org/doi/10.1145/3543174.3546841
https://www.dailymail.co.uk/sciencetech/article-11250965/Jap...
https://videos.dailymail.co.uk/video/mol/2022/09/26/56244997...
British used a LED matrix. More "output" options (frowns and smiles, text), similar results.
https://www.nottingham.ac.uk/news/first-ever-uk-ghost-driver...
https://www.theengineer.co.uk/content/news/pedestrian-study-...
(Not really. But you never know?)
https://cdn.webshopapp.com/shops/70489/files/113563271/600x6...
So if they are showing an ad on the display on your side it means that either it doesn’t see you or thinks you are far enough away that it does not have to account for you in its short term planning.
If it sees you and is taking you into account it could show a diagram showing you and what it plans to do.
I know it sounds harsh, but force feeding me ads I definitely don't want to see while being unable to avoid them (stay home ain't no argument in civilized world) is unacceptable, and there is no but XYZ to make that stance any weaker.
Ads are a cancer of society, not the worst one but the core principle is rotten beyond reason - manipulate general population at your will, against their default choices, for money or power (or worse).
One aspect I'd still like to see them figure out:
As a bike commuter, I share the road with Waymo often. When cycling, you're always checking the eyes of other drivers to confirm if they see you.
I've had a couple incidents where I was a little thrown/unnerved because I couldn't confirm if a Waymo was registering me. It ultimately behaved fine, but I had to put my guard way up because I just couldn't tell.
I've wondered if they could use the light/display on top to signal acknowledgment? How do you simulate the precision of a gaze with a machine that's "seeing" everything at once?