thomas_p at November 11th, 2011 07:10 — #1
this is a bit off-topic but i'm wondering how AI can be used in the future to solve urban problems like increasing traffic congestion. The reason is that I'm working on a research project on smart cities and I thought about an interlinked semantic system in which AI can be used for traffic coordination and the prevention of traffic jams.
I'm also thinking in terms of information sharing between traffic participants to predict traffic flow (which might include thoughts like robotic cars for urban areas). My assumption is that by let's say 2050 we are all so interlinked so that real-time sharing of information is easily possible and so any AI-system can be fed with all the data necessary.
The major question I'm asking myself: is it possible to develop e.g. a semantic system that can analyse all available data (gps positioning etc.) so that it can detect evolving traffic jams and inform all traffic participants who might be affected by that jam? How can AI be used to understand human mobility behavior? If you can think of any solution, just share your thoughts!
stainless at November 11th, 2011 08:37 — #2
Well the first thing you would do is have the ai drive the car and communicate with nearby vehicles via a local bluetooth type system.
Then the cars can drive REALLY close together increasing the capacity of the existing road ways.
Basic traffic jams will not occur as they are usually caused by bad driving, people going from the outside lane to a slip road causing following drivers to brake, causing drivers behind them to brake harder, etc.
alphadog at November 11th, 2011 10:31 — #3
Surely googling "artificial intelligence traffic" will give you more material than anyone could post in this forum? Google itself already has a car that drives itself, even in traffic.
Anyways, first thing that popped into my mind was securing such a system. Oh what a hacker could do with this...
thomas_p at November 13th, 2011 11:34 — #4
Yes but e.g. the google car is only part of the solution because it only cares for itself if you'd like to call it that way. IMO that would reflect an egocentric approach: every car optimizes its own way. But that might not be the optimal way in a system, where all the other cars affect your choice.
There is also the question of ownership: if you own the car, you'd want to park it when it's not in use (thus, it takes away part of the very limited space in cities). If people do not own the vehicles anymore, anybody could use them when needed. This would be like public transport but with a lot more flexibility. Do you think that could be a feasible solution?
Actually I wondered that if there is no central coordination of traffic, it is really hard for any hacker to get into the system. Because if decentralized agents share information with each other in a very flexible way, you'd need to hack every single unit in order to affect the system.
alphadog at November 13th, 2011 15:34 — #5
IMO, if you think of it as "personal rail" instead of "automatic car", you'd get more buy-in faster. Not sure where you are from, but in the US there is a lot of individualism wrapped around cars that would take some serious effort to go against. Salmon swimming upstream comes to mind...
Abu Dhabi has the Masdar system. Take a look at that and other projects in that city.
As for security, you are already going down a common road, pun intended. "Hey, I don't need to worry about it as much, because it's a decentralized system." How do the cars coordinate? Ever hear of a worm virus?
thomas_p at November 15th, 2011 05:09 — #6
I agree, in the US the situation is totally different compared to e.g. Europe. In Europe we have a very efficient public transport system so you could complement the existing system personal rapid transit systems (PRT) where needed (that could be controlled by AI and function like on-demand systems).
In America people strongly depend on their cars, so you would have to provide flexible personal vehicles that take you from home to work (and not only to the next train station etc.). That would also increase costs and thus limit the feasibility of such a project.
Thanks for the Masdar hint, I already know the project and I've also had a look into it. Functions pretty well but it's a totally new city with tremendous up-front investments. I highly doubt that you could just copy their system and introduce it to other existing cities (I mean infrastructure is rigid, you can't just easily replace it).
Security aspect: I didn't mean that about wouldn't worry about it because if security can't be provided you don't need to implement such a system. I just wanted to hear some thoughts on how security could be provided. Nonetheless, IMO a decentralized system is securer than a centralized system
stainless at November 18th, 2011 05:01 — #7
Yes, I agree.
Having a seperate series of small ai's controlling local vehicles is much more secure than having HAL sat somewhere running the entire road network.
However you need to think about how the ai would work as a series of layers
Layer 1 Controlling the vehicle
Layer 2 Predicting traffic flow
Layer 3 Route planning
Layer 4 Network control
Layers 1-3 would be in the car, layer 4 an external ai that controls things like traffic lights.
Layer 2 would handle a car that needs to change lane because it is leaving the current road network. All the cars around it would vary there speed to allow safe passage.
Layer 4 would talk to ai's and advise them of local traffic problems and this would be an input into layer 3
I think it would work really well, but the possibility of a hacker gaining control of the system is a bit scary.
Having said that, in the last 5 days I have spent about 25 hours on british motorways. I think a hacker would probably be safer than some of the drivers I have just seen.
alphadog at November 18th, 2011 09:14 — #8
If the "small ai's controlling local vehicles" are communicating with each other to coordinate traffic, then you essentially have to treat the system as one. IOW, if I capture one node, I p0wn the system.
Actually, assuming you can create a secure way to note vehicles, it would not be hard to compute the best traffic situation.
The primary problem I can see is that what may be best overall may not be best for me. If the best way to keep traffic to a minimum is to route me out of town then back in, I'd be plenty pissed...
stainless at November 29th, 2011 07:00 — #9
LOL yes, I have played around with sat nav's and it is possible to create some amazing routes
I managed to get one route which I flagged as "avoid motorways" "fastest route" from Southampton to Liverpool
The route was Southampton to Bristol, ferry to Ireland, up the east coast of Ireland, ferry to Liverpool