Hacker News Reader

Comment by tbabb | Hacker News Reader

I wonder how city congestion will be affected by thousands of empty cars driving around a few years from now.

It would actually reduce congestion. When I had a car here in SF, in many cases, I spent more time driving around looking for parking than actually driving to and from my destination. This would prevent that since the car could drop you off and head back home or to a nearby area with more readily available parking.

Take this idea to the next level, and you can have services like Uber that are run completely by algorithm. These cars can be parked all around the city and as soon as someone requests an Uber, the car wakes up and picks the customer up, after dropoff it finds the most efficient place to repark itself and wait for the next customer. This would significantly reduce congestion in big cities.

How the heck would the car know where there is more readily available parking? It seems like parking is a whole other area of work that needs to be done. It would need to: - Know local parking lots - Know that if a sign says "full" it can't go in - If it's not full, somehow grab the ticket out of the machine - Know the difference between parallel parking and other types - Know if a meter is expired - Pay for a meter if it is, or not if it isn't - Tell how much time is left on a meter - Be able to read parking restriction signs (street sweeping, restricted time ranges) - Etc. etc.

Edit: tried hard to get the above to output as a bullet list, couldn't get it to work.

Cloud parking. If a car spots a parking spot (cities could encode these in Google Maps, too, for example) which is free, it notifies the cloud.

Cities may also take advantage of already ubiquitous CCTVs and optical recognition software to provide parking info.

parking garages/lots would be incentivized to work with manufacturers to come up with automated ways to check in and out because it directly leads to more revenue for them (and less hassle with accidents I'd bet)

Just availability is already available and popular in some cities, Seattle for instance http://www.seattle.gov/transportation/epark/default.htm

I think you're vastly overstating the complexity of figuring if a parking spot is available and paying for it.

a. You can use parking / traffic data to find neighborhoods most likely to have available street parking and look for parking there. As for solving the problem of when you're allowed to park somewhere, this is already being worked on [0][1] and I would put money on it being ready before autonomous cars.

b. Use parking garage app to find the best balance between distance and price of parking for the time needed. As far as determining if the garage is full, many garages have sensors now that count how many open spaces there are. It would not be stretch to assume that by the time autonomous cars are ready, this data will be available within the parking apps.

c. One of the cool things about having a computer make these decisions is that it can factor in a limitless amount of data to make the most efficient decision. Kind of how Google Maps gives you different transportation options when you ask for directions, you could do the same with parking. The car can ask you, which of these options would you like:

1. Drive 30 minutes in this direction and you can park for free. Cost for power to drive there: $1.12 - Time to summon: 30 minutes - Cost for parking: $0.00

2. Park in a parking lot around the block: Cost for power to drive there: $0.08 - Time to summon: 5 minutes - Cost for parking: $10.00

etc...

I agree with you though, parking is a huge issue in the city and I really can't believe it hasn't been fixed yet. We are getting there, but very slowly.

[0] http://sfpark.org [1] http://spotangels.com

I wonder how long before New York City walkers realize they can safely cross in front of Teslas without worrying about getting hit.

Do you think dash cams + automated flagging of such incidents by the software could counter that?

Counter what? Should people not be allowed to cross the road?!

No, not that. The comment I replied to was implying that once people figured out car AI wouldn't kill them, they'd just halt cars for the fun of it or when they felt like it.

With a fallible human behind the wheel, you can't take chances like that.

(To be clear, I don't drive and am for pedestrianism. But I can see how it would be a problem if people can simply block autonomous cars. I'm also worried that a heavy handed response to that problem could harm pedestrianism.)

I've been wondering, when we get to the point that there is a mix of autonomous and human driven vehicles on the road, if we will see the human driven vehicles start to game the autonomous vehicles' collision avoidance systems? For example, the main reason not to cut off traffic is because the driver might not react quickly enough (ok, the other reason is no to be a jerk.) but with autonomous vehicles, human driver may be able to become significantly more aggressive, knowing that the autonomous vehicle will get out of the way.

(As a side note, I wonder how autonomous vehicles will handle lane splitting motorcycles? They get pretty close to the cars to either side.)

The autonomous car could automatically send the video and sensor data of illegal human driving behavior to the appropriate authorities. It seems pretty stupid to drive dangerously around a car that's recording so much information.

Google has a patent on algorithms for detecting dangerous drivers:

https://www.google.com/patents/US20110202305

Free idea I've kicked around: Give away free dashcams to drivers that upload their footage nightly to Google (with user's permission), have Google process the data to find high risk drivers, sell that information to their insurance companies to properly re-price their premiums. Take a cut, share said cut with dashcam drivers.

Drive the price of human driven vehicle insurance high enough to where self-driving vehicles are the logical choice.

For drunk driving and near accidents? Sure. For tattletale speeding tickets? No thank you.

Did you just suggest that these drones should be programmed to record our activities and report them to the police?

A self-driving Uber-style limo/rideshare service might make sense and may result into less cars overall.

It would certainly free up a lot of land which is currently being wasted on parking lots. I wonder if there will be a time when most people don't own a car, but instead they use a rideshare like that.

Self-driving Ubers need cleaning and ensuring safety. Without cabbies to take care of that, I think that's a major reason owning a car still has value - it's a private space.

I've read estimates that 30% of city traffic is made up of people looking for a parking spot close to their destination.

Reply on news.ycombinator.com