I am just wondering how many people have to die before it finally dawns on on us that the current approach is idiotic.
The supposed smart cars can't even talk to each-other directly 'without internet' to warn of a car pileup! They can't report to city traffic control about their condition.
We should be innovating in infrastructure instead - create standardised computer readable infrared road markings, equip each traffic light and each lamp post with a radio beacon, each crash barrier could have a radio marker, create PUBLIC maps of each city, have a central traffic control sypercomputer in each city provide directions to cars. Have each car painted with infrared markers so they recognise each-other. Provide cyclists with something these cars can recognise.
We could even make radar-reflective pants so that autonomous cars see them better.
the whoe traffic system needs to be looked at and brought to a new set of standards, whatever they may be. I am not sure what they are, but it should be clear to anyome with half a brain thay having a car use a camera to tell if the traffic light is red or green is idiotic.
untill a new system is ready, no car without AGI level ai will ever be safe
The only problem is that such collective approach conflicts with the way VCs work.
40k people are dying a year from cars, and the problem is that driverless electric cars are not the solution. They are the faster horse kind of innovation. If we were serious about reducing fatalities, improving health, and reducing pollution, the solution would be to drastically reduce the amount of miles driven per year and the number and the size of the cars. We need to encourage people to bike and walk and use public transportation. Autonomous trains would work extremely well and we don't have to worry about bicyclists. The zoning laws and car centric development of cities is the issue.
> The zoning laws and car centric development of cities is the issue.
Well, the main issue is that most of the cars on the road during the busiest hours have just 1 person in them.
I think when the dust finally settles, the main form of passenger transit in urban-ish areas will basically be something that looks like autonomous vans.
Unlike public transit today, they won't have predefined timetables, nor predefined startpoints and endpoints. They'll pick you up wherever you are, whenever you want, and drop you off wherever you want, but they might take a roundabout way to get there to pick up and drop off other passengers too.
If they usually have 2+ people in them at a time, we'll be able to use our existing infrastructure (roads) at a much higher throughout.
How is this an improvement over trains, trams, and light rail?
I think reusing car-based infrastructure with busses is an important step in transitioning away from car dependency, but roads are not permanent. In many parts of America, trams and railways were torn down to be replaced by roads. The roads can be replaced too, if it is clear that alternatives are better.
Plus, roads are destroyed incredibly quickly by cars and busses, so they will need replacing eventually, whereas rail is much more resilient.
It seems to me that rail-based transport is much more effective in dense urban environments.
Point to point transit at the time when you need it.
If I want to get from A to B at 14:00 and can do that in 30 minutes by car, but the only bus that goes there requires me to change at C, and it takes 20 minutes to get from A to C, 20 minutes to get from C to B, a 10 minute wait time at C, and the connection happens every 30 minutes and isn't perfectly aligned with my schedule, I'm now facing the choice between 30 min by car or over one hour (including the wait time at my destination because I had to take an earlier connection) with public transit. Add a 5-10 minute walk at each end to account for the distance between the nearest stop and the start/destination, and public transit can take 3x as long.
If there is a van service that gets me there in 40 minutes at a time of my choice, that's a much more acceptable option.
This may also consist of a car first taking me to a tram/bus stop a few minutes away that gets me more directly to the destination.
The van service also solves the problem of those few-percent edge cases where regular public transit just doesn't work, which force people to get a car, which then gets used even for trips where public transit would otherwise be acceptable (but not great).
This is already possible today. You don't need driverless cars to do it. Uber can already do it, but it's not taking over. Unless you're suggesting we will have more cars just sitting around waiting for rides, because they're driverless. Then we get back to them being so convenient that nobody will want to ride with another person.
In other words, why does making the car driverless suddenly make everybody want to use it?
So they if get rid of the drivers they contract who is buying and paying for the cars and maintenance? - are you suggesting Uber start carjacking their existing drivers?
Do you know what percentage of driver compensation effectively goes to car ownership and maintenance?
None or few of those options are particularly accessible (EDIT: for people who cannot walk/cycle long distances). Public transportation often is quite inaccessible while still claiming to be accessible (like the London Underground and the New York Subway). Central management of driverless cars could solve the problem in a more inclusive fashion and isn't just more of the same idea. We should still reduce the size of the vehicles, though.
EDIT clarification: Accessibility for wheelchair users, blind people, and anyone with ambulatory restrictions. Most stations are not wheelchair accessible to the train and having to go on much longer routes to compensate is not equal access. And many 'encouraging other use' initiatives block all cars, leaving these people in the lurch. Something that gave door-to-door access to everyone, equally, via a fleet of cars, could be superior.
If we could reduce car infrastructure to the point where only people who need a car for the transportation to be accessible would use it that would be a dream.
I don’t get the accessibility angle at all. It’s a non-issue because you will have a very hard time finding people who have issues with exceptions for accessibility. That would still reduce the need for car infrastructure massively.
The thing is, we need to start building the future right now, right here and massively. That means public transportation centric development (= building public transport to nowhere and letting the area develop alongside the walksheds with appropriate mixed use zoning) and actually being brave enough to re-structure existing solutions along those lines. That all isn’t magic, we just need to do it. We know it works, it doesn’t require any technological breakthroughs.
>> None or few of those options are particularly accessible.
At the moment, neither are self-driving cars sufficiently capable to replace human drivers.
We're chasing a future tech dream when we already have technology that can solve the problems we are facing right now.
Edit: What do you mean by "accessible"? The London underground and generally public transport in the UK works very well. Speaking after having lived there for 15 years, studying and working in London and a smaller town, and never having owed a car.
I can tell you've never tried to use the Underground with a wheelchair, tho if you're questioning what accessible means in that context—because it's really bad. cars are very accessible for those with mobility disabilities & for many others.
I don't use a wheelchair and I don't know about the London underground but I can tell you that public transport in the small town I lived for 15 years were eminently accessible, as evidenced by the continuous presense of passengers with obvious (and varied) mobility issues on them.
In any case improving accessibility on public transport also does not require any new technology that we don't know how to create. It just requires political will.
This is a common refrain, that things must be accessible because you see people with mobility issues using them. But it's not true - the people you see are the people who are coping with it, either all the time or on that day. You don't see all the people for whom it is not accessible. It's well-known in disabled circles that most public transport is hit and miss at best and exclusionary/ableist/injuring at worst.
Actually solving the problem does involve new technology, because the worst cases need door-to-door access and most cannot afford taxis (nor do welfare systems provide them).
What makes it all harder though is the continual fight to persuade every last person that it is a significant problem, because they're so sure it can't be.
Just going with the parent comment and I'm asking when did you last use the Underground and what station do you think has accessibility issues? I found that the London system is pretty much accessible with aggressive retrofitting on older stations, but I'll acknowledge that if you have used it say more than 10 years ago then I understand.
I've used it plenty and still use it. It is getting better in places, but just look at the tube map - large parts of the system still lack step-free access, let alone the problems the system has in other ways (overcrowding and people not giving up their seats or making space appropriately); we should be trying for (in-city) transport systems that don't require such congregation.
The last time I travelled the tube with a wheelchair user with a pain condition, the (non-strike) closures hit in such a way that we couldn't reach our destination without going on a detour that was so long (more than an hour) that the system assumed we'd forgotten to swipe out and back in and charged a penalty fare. The people manning the help-intercoms were unable to assist in finding a route that didn't have closures. And it's normal to have to find alternative, slower routes than the ones everyone else can take.
+1 on London underground being very accessible. Although I only lived there for a few months, it was insanely easy to get around relative to American cities I've lived in (San Diego, SF, DC) although I think Berlin's combination of train and tram was even better
Surely it would be easier to make public transportation more accessible in a non-car oriented city than making driverless cars, which then have large amounts of continuous infrastructure costs, health care costs from less exercise, social costs, and also the fact its still not known if this is even entirely possible.
As a thought experiment, we have no cars other than emergency vehicles and some commercial vehicles occasionally. Everyone is able to walk to 90% of destinations, and bike or public transport the last 10%. It would be easier to have people who can't walk well in small electric personal transport devices (wheel chairs, segway, electric bikes, motorized scooters seen at walmart) that if they crash do not pose any threat to anyone. Surely adding wheel chair accessible elevators to train stops for public transit, is easier than building out entire car centris highway infrastructure that requires continuous expensive maintenance. Public transit and walking is already more accessible to the blind than automobiles, that is a nil point.
Or 1m+ people if you count non Americans as people. I'm optimistic that self driving tech will be able to be repurposed to prevent accidents even if fitted to non electric non self driving vehicles. The switching to bikes etc. is a nice idea but probably won't happen in a hurry.
This seems to be the dominant opinion is here, that is, to approach this problem as we approached civil aviation and flying commercial planes.
To respond in a sentence: first, you have no proof that "investing in infrastructure" would actually yield working autonomous cars. Second, there are no proof either that the current approaches are doomed to fail.
To respond in details: everything that you propose doesn't seem to be able to solve the main challenges that self-driving cars are facing. These challenges are all about edge-cases, and responding safely to changing conditions and misbehaviors from other agents.
The response cannot be "let's add a ton of fragile electronics in all of the infrastructure" because this: will cost a ton of money to build and maintain, will break in weird ways over time and lead to cascading failures, and most importantly: it's dubious that it would even work at all.
The final comment about VCs is weird: a lot of actors in this field are not VC backed startups.
My opinion is: Tesla might have a working model that outperform human drivers in 10 years (50% confidence)
> These challenges are all about edge-cases, and responding safely to changing conditions and misbehaviors from other agents... The response cannot be "let's add a ton of fragile electronics"
Self driving car has failed to detect a person in the dark and killed her. It was demonstrated that Tesla autopilot fails to detect a child crossing the road. It confused a semi trailer and an underpass. You call that an edge case?
> The response cannot be "let's add a ton of fragile electronics in all of the infrastructure"
So instead we will add a ton of fragile electronics in every car. Instead of having 3 optimally positioned lidars monitoring an intersection, we will have 20 poorly positioned lidars on every car.
You don't need electronis, you can use paint. You can put QR codes that tell a car where it is in an underground parking lot. If we add a parrern to all clothing in infrared-only paint, that would help various AI detect people and tell them apart from surroundings. You can put a metal patterns on street signs that are detected by automotive radar.
> will cost a ton of money to build and maintain
And electronics in the cars will cost even more
> will break in weird ways over time and lead to cascading failures
You mean like the failure the main article is discussing right now?
> it's dubious that it would even work at all
And it's dubious this 'individual' approach will ever work. Every single argument you've made equally applies to the approach you are advocating.
Sounds like a lot of infra and related spend via taxes or muni debt for a tech outcome that’s a privacy and autonomy nightmare.
Build some trains, build some bike lanes, and leave the people alone who don’t want their transportation and the downstream reasons people use it taken over by tech.
Governance by a central cybernetic super computer and all the cybernetic hypothesis coming true. Good lord.
Indeed. I raise issue with the unthinking advocacy here, as if civil society is just a canvas for engineers to paint over. It’s the only thing I hate about this industry. It’s UX-oriented as long as tech is part of the UX.
I may sound like a Luddite but I recall being able to go out in NYC without needing to QR-code scan my way through daily life.
I think you hit the nail on the head with trains, because your parent is pretty much describing a modern train signaling system, down to the centralized supercomputer.
> create standardised computer readable infrared road markings, equip each traffic light and each lamp post with a radio beacon, each crash barrier could have a radio marker, create PUBLIC maps of each city, have a central traffic control sypercomputer in each city provide directions to cars. Have each car painted with infrared markers so they recognise each-other.
It is amazing how often tech savy folks can reinvent trains.
I think eventually major car manufacturers will work with the regulator to bring on new standards, and it will happen withing 15 years. Infrastructure upgrades will take another 15. Most startups that aim to solve self driving will either find a niche in that ecosystem or go extinct.
So I expect reasonable self driving cars to be common in depeloped nations in 2050, while AGI will be just as elusive.
In a similar vein, the correct solution to electric trucks is not giant batteries, it's simply overhead wires:
>>The supposed smart cars can't even talk to each-other directly 'without internet' to warn of a car pileup! They can't report to city traffic control about their condition.
- This by far in my opinion is the single most important attribute of any autonomous fleet of vehicles that is needed. There should be first of all a standard of communication via RFC defined for autonomous vehicles and drones. Any vendor manufacturing autonomous vehicles or drones should follow this standard of communication and be able to communicate to each other with the primary objective of avoiding collisions. This should be a RFC so that it works across vendors and is as standardized as TCP/IP. Communication medium can be over radio frequencies as well as the internet and can work as backup of each other but first there should be a established standard that all vendors adheres to.
I think this is a bad idea. The potential for a malicious/incompetent/malfunctioning actor to wreck havoc becomes enormous.
Vehicles should assess and plan driving from first principles and physical sensory inputs. Networked inputs could be helpful as augmentation but it's not something one should rely on at all. Kind of like you shouldn't take a human showing/not showing a turn for a reliable signal of intent.
Yes this is augmentation only but it is essential to have. This is what replaces a driver to driver communication part. Like two drivers at a four way crossing sometimes signaling each other.
BTW being opinionated is fine but no need to downvote
Seems to me you’ve just reinvented a city rail transportation system with modern signalling. Modern rail system can indeed drive fully autonomously if they are fully grade separated from street traffic.
That's a lot of effort, money, and infrastructure that could go into developing a healthy public transportation system instead of trying to stop all the cars from killing people.
You're describing exactly the kind of autonomous car development that was going on for the last 40 years. It went absolutely nowhere. It wasn't until tech got good enough and developers started treating vehicles as independent agents that any real progress has been made.
The supposed smart cars can't even talk to each-other directly 'without internet' to warn of a car pileup! They can't report to city traffic control about their condition.
We should be innovating in infrastructure instead - create standardised computer readable infrared road markings, equip each traffic light and each lamp post with a radio beacon, each crash barrier could have a radio marker, create PUBLIC maps of each city, have a central traffic control sypercomputer in each city provide directions to cars. Have each car painted with infrared markers so they recognise each-other. Provide cyclists with something these cars can recognise.
We could even make radar-reflective pants so that autonomous cars see them better.
the whoe traffic system needs to be looked at and brought to a new set of standards, whatever they may be. I am not sure what they are, but it should be clear to anyome with half a brain thay having a car use a camera to tell if the traffic light is red or green is idiotic.
untill a new system is ready, no car without AGI level ai will ever be safe
The only problem is that such collective approach conflicts with the way VCs work.