5 ms·
Car Autonomously Powerslides Into Parking Space
- dan00 16y agoIn the future you have to write the better algorithm than your contender, to get the parking space.
- roc 16y agoIn that future my car just dropped me off at the door and is going to park itself. Who cares if it finds a close spot? Whenever it does find one, it can tell me how much notice it needs to pick me back up. The difference in wait between today's "good" and "bad" spots will be a couple minutes at most. Non-issue. More interesting will be regulations regarding drop-off/pickup-points and how much enforcement the legislation requires code to do. e.g. can not allow double-park. can not idle for automated pickup for longer than X-minutes. must download list of acceptable parking zones by date. must report every automated parking location/duration. etc
- mattmaroon 16y agoIn the future there won't be cars, we'll just beam ourselves anywhere we want to go. If we're going to dream, might as well dream big.
- nooneelse 16y agoAhh someone who has thought past the surface, "can do what we currently do better" of this to some of the more disruptive changes that autonomous car technology might bring. The use of the car's logs to meter all parking in a city is nice. Some new gaming of systems will begin. The price of parking gets weighed against the cost of just driving the car in circles for the duration of the stay, and the defector strategies, like shifting the car between locations which have a "cars can wait X minutes for free to do a pickup" rule, need to be watched for. A business model that might well feel the mixed bite of autonomous vehicles is roadside hotel/motels. If you invest in an autonomous car with "sleeper options" (or rent a sleeper car, bed(s) on smart wheels), any town within about 8 hours drive becomes psychologically much closer to home, just a night's sleep away. It becomes closer to absurd to ever stop at any hotel between you and your destination or to take such trips so that they use your waking time. But then hotels at the destination can cater to the "just want a safe parking area and a shower/bathroom in the morning" crowd. Opportunities here too for the car rental companies to provide end-to-end "sleeper car solutions"; competition and/or cross-promotion between car rental companies and hotels/parking-and-bathrooms chains. Towns geographically central between large farther-flung working cities could become hubs for meetings, think Voronoi diagram. Or they can choose, instead, to cater to the weekend tourism crowds from those large cities. Somewhere around Eureka California will become the easy weekend meet-up mecca for practically the whole west coast.
- mhb 16y agoSounds a lot like a comfortable bus. And where are those?
- nooneelse 16y agoFair point, but it has a couple of differences which line up with the reasons people give for not using buses, trains, and planes. Those are: no need to fit the travel plan into the bus schedule and having a vehicle in the destination city for no extra charge. People aren't always right when asked about their own reasons for choosing one option over another, but I think in this case these are important motivators as they add to the cost in terms of both money and also the psychological "hassle" factor. Further there are social signals sent by how one travels, and buses have let themselves become, or targeted being, the option for low class travel (unless it is one's personal bus, but even then it is still below the personal plane). Autonomous cars would have the opportunity to aim their marketing toward them being seen as a desirable, upper middle class travel option (or the un-extravagant option for the more wealthy). The cars themselves will likely be more expensive than cars with no autonomous capabilities, so they easily send a signal of "well-to-do".
- mmphosis 16y agoOkay so my car Autonomously Powerslides Into a very tight Parking Space. Good luck getting out of that same parking space.
- KoZeN 16y agoIt won't be long before traditional grease-monkey mechanics are replaced by code-monkeys. The advances in automotive technology within the last three to five years have been staggering. I'm genuinely looking forward to seeing what cars my kids will be driving, or not, as is the case here!
- thefool 16y agoyeah, but they are now no longer really tinker-able, or even fixable, because so much of their operation cannot be discerned. So when they do break the headaches are much bigger.
- michaelkeenan 16y agoIf you'd like to learn more about how autonomous cars will change the world, I strongly recommend the robocar essays of Brad Templeton: http://www.templetons.com/brad/robocars/ http://www.templetons.com/brad/robocars/ Highlights: -Cars that don't crash due to human error would save 40,000 lives and a million injuries (NIH). Mostly young people, for whom car accidents are the leading cause of death among major categories. -Not crashing would save 230 billion dollars of accident cost (NTSB). About 2-3% of GDP. -Not driving would free up 50 billion hours (or 1 trillion dollars) of people's time. Around 8% of GDP. -Electric cars (more feasible when the car can go to a service station when it's not being used, because one of the major problems with electric cars is that storing lots of energy in a battery is difficult) would save 50 billion gallons of imported gasoline, replaced with the equivalent of 5 billion "gallons" of domestic-source power plant fuel. Thus eliminating about 12-15% of the USA's CO2 emissions and nastier pollution. -There would be a serious reduction in the urban land devoted to the ~600 million parking spaces, estimated to be up to 10% of urban land in many cities.
- aidenn0 16y agoI thought about this though and there is a serious problem. Let's say I've created a robo car system that is 100x less likely to be involved in an at-fault accident. If everyone installed the system nationwide, I would be assuming about 2.3 billion dollars (1% of the accident cost figure quoted above) in liability. However, that's not likely to happen since the first time such a system fails and crashes it would be horrible publicity that would put me out of business.
- jasonkester 16y agoBefore giving these students an A, I'm going to have to ask that they replace the cones with their own cars. That's where you draw the line between "I think this will work" and "I have complete confidence that this will work". It's analogous to if I were a welder working inside a giant propane tank. The guy with the gas meter can swear up and down that it's safe to fire up the torch, but I'm still going to insist he stand in there beside me when I spark up. Besides, it just makes better video. It's not anywhere near as impressive if there's nothing on the line.
- bad_alloc 16y agoEspecially they confidently tell us, that the system is accurate to two feet. Sadly two feet is the difference between parking ok and damaging both cars significantly (i.e. you'll have to at least fix your bumpers.)
- dugmartin 16y agoIf you look at the tire marks on the ground it looks like they ran this test quite a few times. I wonder if they ended up just moving the cones where the car happened to stop and then took the video.
- maukdaddy 16y agoAfter what seems like 10 clicks I finally found the source: http://spectrum.ieee.org/automaton/green-tech/advanced-cars/autonomous-car-learns-to-powerlslide-into-parking-spot http://spectrum.ieee.org/automaton/green-tech/advanced-cars/...
- mattmaroon 16y agoDoes it autonomously change the tires when they go bald?
- briancooley 16y agoThe application (parking a car by powersliding) is a nice sizzle for a really interesting (to me at least) steak. The idea of combining closed loop model-based control for the easy-to-model portion of a process with open-loop control for the difficult-to-model portion of a process would have a lot of application in lots of places, but not many people would pay attention to using the technique for, say, controlling a distillation column. I appreciate the nice link to a PDF I found on Mr. Kolter's website from the link at the end of the second video: http://ai.stanford.edu/~kolter/lib/exe/fetch.php?media=pubs:kolter-icra10.pdf http://ai.stanford.edu/~kolter/lib/exe/fetch.php?media=pubs:...