Given that these vehicles will shift the risk model from driver to manufacturer, and subsequently programmed to obey the laws (which include right of way), we might actually see pedestrians and cyclists using their right of way instead of being bullied away from it.
The idea that cars can communicate with each other so they could drive closer together and faster is complete bullshit...Sure, it's possible, but what profit-driven company would ever take that risk knowing the real-world reliability of wireless communication?
The idea that they won't drive around looking for parking? Welcome to your traffic jam of the future: https://twitter.com/yann_rouen/status/807781862022246401
Pretty much all the evidence points to very slow traffic in the future of self driving cars. As someone who mostly walks everywhere, I'm pretty excited.
It may not make the traffic in downtown SF faster, but it sure will make the traffic on the 101 from SF to San Jose faster. And good luck walking that.
And, to build on this point, a lot of urban ones are produced or exacerbated by improper behavior at stoplights.
The shock wave phenomenon is one caused by physics (it literally is modeled by fluid dynamics simulation), exacerbated by limited visibility, and exacerbated once again by human reaction times. You can only get rid of the human reaction times, and maybe a little bit of visibility due to communication (although I'm extremely skeptical that there is any incentive for cars to rely on wireless communication to make decisions).
Lanes merging is a physical bottleneck. Sure, humans might make the merge worse, but speed is still limited by the capacity bottleneck, not the friction of the merge. The zipper merge has never proven to be faster, merely safer and more space efficient. You'll get to the bottleneck faster, and cars that try to exit before the bottleneck will get out of your way faster, but that's about it.
Second, and more importantly, when you have only autonomous vehicles on the road, you can make assumptions that all drivers are perfectly rational, and then your safety margins can be smaller, allowing closer travel at higher speeds.
I suspect in the future, much like how when we transitioned from horse drawn carriage to motor vehicle, at first they will share roads, but then there will be autonomous only roads where the speeds will be higher.
Are you saying that a rational driver will never apply maximum braking?
Because there's a huge difference between individual agents trying to maximize their own benefit to the detriment of everyone else (while also operating based on limited, local sensor data, and with limited processing power and slow reaction times), and a fully integrated traffic flow where fast, powerful agents communicate with each other, share sensor data and telemetry, and are able to make global optimizations.
Scenario #1 often produces pathologic outcomes that are way below global optimum. Scenario #2 can optimize traffic to the extent that human drivers would be unable to keep up with it (think rush hour traffic moving at legal speed limit nearly bumper to bumper with no accidents).
If the kids being dropped off at that school started using self-driving Uber Pools, then that video would be better described as "Welcome to your traffic jam of the past."
I'm open to the idea that carpooling might increase, but where will it come from? If it is coming from public transit, that would be a net increase in congestion. If it is coming from cars, it would be a net win...but self-driving cars should make ubers cheaper, not more expensive, and the relationship with supply/demand at lower cost would suggest that people would move away from uber pools and towards uberx. I would imagine that the more flexible ride-matching of uber pool (compared to traditional carpooling) would make it easier to use uberpool than it currently is to carpool, and with the right incentives (congestion charges that are waived for carpooling?), we could probably make it work pretty well. I don't know though, the complexity of the dynamics here makes it pretty hard for me to predict with any confidence.
I think it can be done; the following car may have to brake hard if communication is lost with the leading car.
Let z1 be the safe following distance behind a non-smart car, and z2 be the safe following distance behind a smart car. z2 < z1.
Follower approaches leader and settles at z1. Follower attempts to contact leader. If contact is successfull, follower measures link quality and computes z3=f(z1, z2, quality). If link quality is perfect, z3 can be z1 - in which case the slightest link disturbance causes hard braking.
If link quality is 50%, z3 might be avg(z1, z2), and the follower has more time budget to avoid braking during tiny dropouts.
because that's not the most efficient way to move traffic. Yes, a road at it's limit is going to be worse off. Perfect merging and other better habits will help in a lot of places. Just think of all the times you see some idiot merge in at 50 way before the end of the lane.
Their PR department are, as can be expected, top notch though. I especially like how they put a populist spin on their announcements, like the beer delivery (yay, beer!) and now picking up passengers with their proof-of-concept vehicles, to make it look like self-driving cars are already part of their business.
Too many tech companies (and esp. robotics ones) do their development and testing under limited conditions, and then they're blindsided by real-world realities.
Getting contact with customers early is key. It's like the old military adage: No battle plan ever survives first contact with the enemy. In this case, the "enemy" is the real world.
Kudos to Uber for getting out there and working with customers -- it's much better than Google's autonomous cars, which have driven millions of miles and never had a single paying passenger!
Feeling peckish? Ignore it. It's just the McDonald's corporation trying to raise the value of it's stock.
The prices are lower because they're subsidized with invested money and debt.
One thing much of this analysis is missing is how much more in demand commuter vans will be. Transportation will be much more hub based for both long and very short trips. A middle of the range option will fill the void, which is large. Given much of transit is pre planned with times and locations, for jobs, van use in self driving cars will be enormous.
Edit: It should be noted this was in the context of a discussion regarding ad-hoc pairing systems for carpooling/ride-sharing possibly making car travel much more efficient, possibly with vehicles that seat more (minivans, full size vans, small busses).
---
I actually wonder about this. Is it a matter of most cars being mostly empty, or is it inherently impossible to match the mass transit capacity. If we look at it as passengers carried over space required and time spent, here's how I see it:
Rail mass transit does not as efficiently use the land it's on a occupancy basis (there's not always a train on a specific square foot of track). In peak times, cars are more efficient on a vehicle basis. According to BART system facts[1], there are 107 miles of track. There are 669 cars, seating for 72 in 448 of them with each being 70 feet (with 59 of them having an additional 5 feet for a cab), and seating for 64 in 230 cars[2] (with an indeterminate car length, so I'll use the smaller listed), for a total capacity of 46,976 seated people. BART states that all cars can hold over 200 people in a "crush" load, so we'll assume 200 as the theoretical maximum, and say BART can carry 133,000 people when at peak (crush) capacity, and over the 107 miles of track, that gives us a density of 1,250 people per linear track mile, but with only 8.3% track utilization at any one time.
Cars do not as efficiently pack people per vehicle usually, but can more efficiently use the roads on a per-vehicle basis. Assuming very heavy traffic which is not stop-and-go, so perhaps 35 miles an hour average (the same as BART), and a 4-lane highway (two each direction), if each car is allowing two car lengths between itself and the car in front (slow traffic), we have approximately 33% road utilization (or 25%, or 20% depending on what you think the average space between vehicles is). Since carpooling seems to be at about 10% currently carpooling[4] (ignoring that it may be different in certain arterial routes, as we are discussing), we have around 1.066 people per car[5] as a lower bound. With an average car length of 177.2 inches[6], or 14.77 feet, we can estimate the people per mile on the highway during this time as being 604 people per mile of 4-lane highway.
Interesting take-aways for me:
While 4-lane highways may take more room than rail (not sure the actual sizes here), they are also more versatile.
If the highway bogs down below 35 mph, it's then less than the average rate of BART, and we need to start computing people over time instead of just people over distance.
BART has much more room to increase track utilization, but there is likely unaccounted for overhead here on each train. Optimal usage at current speed is one train arriving immediately after the prior one leaving, at 35 mph exact speed and 20 second stops, for a train of six cars (?) and 425 feet, that would be cars 2.42 train lengths apart, and a utilization of 29%, or roughly a 4x increase over current rates, 5,000 people per track-mile.
Cars have much more room to increase vehicle utilization. If we replaced 50% of vehicles with full size vans transporting on average 7 people each and didn't touch road utilization, we would be at an overall average of 4 people per car, and 2600 people per highway mile. Interestingly, if we somehow moved towards a system where smaller vehicles picked up and shuttled people with small amounts of sharing to bus-stations where they were sorted into smaller buses (40 people) going specific area depots, and from those depots dispersed again to final destinations using individual cars with small amounts of sharing, we might easily surpass rail transit systems. averaging 20.5 people per vehicle, but with somewhat more area used should put us close to 10,000 people per highway mile.
Of course, there's a lot of assumptions in all the numbers, and some speculation in the possibilities, but I thought it was interesting to figure out. ;)
1: http://www.bart.gov/about/history/facts
2: I know the total car numbers don't add up. Complain to BART, it's their data.
3: (66970 feet + 695 feet)/(7 * miles * 5280 feet/mile) = (47125 feet)/(564960 feet) = 8.3%
4: https://www.census.gov/prod/2011pubs/acs-15.pdf, table 1.
5: 105,476 drove alone, 13,917 carpooled, if we assume all carpooling was just two people per car, we get person to car density by (105,476+13,917/2)/105,476 = 1.066 people per car
6: https://www.reference.com/vehicles/average-length-car-2e8538....
7: 5280 feet/mile / 14.77 feet/car * 0.33 highway mile utilization = 117 cars at highway lane mile utilization. 117 cars * 1.066 people/car = 126 people per mile of highway lane. 4 lanes fives us 604 people per highway.
And (as with the "smart home" advocates), they forget that, outside of the tech crowd - not everyone likes interacting with (and hence, being dependent on) computers for every conceivable need in their life, 24x7.
The ownerless model might work for some people -- but not too many, I suspect.
Sort of like how you couldn't open your own payment processor for the few hundred people in your neighborhood today.
Worldwide, we are at ~54% living in cities. [3]
This trend seems to be continuing globally as well [4], so pretty soon people in non-urban environments may have no choice (assuming, as I do, that cars will get more expensive as they have more and better technology and are much more highly utilized due to sharing)
[1] http://www.census.gov/newsroom/press-releases/2015/cb15-33.h... [2] http://www.citylab.com/housing/2012/03/us-urban-population-w... [3] http://www.un.org/en/development/desa/news/population/world-... [4] https://esa.un.org/unpd/wup/publications/files/wup2014-highl... (page 7, Figure 2)
I have a friend working for a major supplier in self-driving vehicle sensor tech who went to Pittsburg to learn from Uber. I imagine they at least did their homework before flying him out there.
Unlike a Tenderloin sidewalk, deterrents are easily enforced under a scheme like this.
Uber and Lyft currently charge $100 for a cleaning fee to the driver when their passengers soil a car's interior with vomit, etc. I don't see why Waymo, Tesla, or Uber would be any different.
1 in 3... you must be from San Francisco?
If you indicate it is too dirty to ride, the car reports for cleaning. If the cleaners agree with your assessment then your ride is free, paid for by the person responsible for the mess, who would also pay for the cleaning. If the cleaners don't agree, then you pay for the cleaning and your own ride. If you indicate that the car is dirty but want to ride it regardless, then the car could just check in for cleaning after your ride at no expense to you.
If you don't want to send the car away, perhaps you have an urgent appointment, you could just ride the messy car and have it cleaned after your ride. If adding a few extra minutes to your trip isn't a big deal, then the reward of having it be free may compensate you for the inconvenience.
A system like this would provide an incentive for customers to report dirty cars (saving money on their ride) and to leave cars clean (avoid paying for car cleaning). Those incentives are hopefully at least as strong as the incentives in place now.
(1) Joke's on you. We already know better.
A few years time to market. Damn. We may get accurate sickness data worldwide before we get driving cars.
http://www.economist.com/blogs/gulliver/2009/10/what_price_v...
"I'm a vomit-bucket-half-full sort of guy. I don't think these cabbies are trying to charge you for puking, I think they are offering you the premium service of vomiting in their taxi."
"America was built on the idea of premium services. This is how the wealthy are able to have so much more fun than everyone else. They can behave however they want as long as they have the money to cover the premiums! The $70 currently in my wallet entitles me to a good three blocks on Michigan Avenue and a nice half-digested deep dish pizza projectile vomited all over the headrest."
http://uberpeople.net/threads/throw-up.630/
"Yep Uber is pretty good here in Sydney as well. I had the outside of my car given the Jackson Pollock chunky rainbow look by a passenger. His friends thought it was a real laugh and afterwards I found they had also taken my giveaway chocolates as well. Sent the report in with photos and $95 receipt for the cleanup, and got $250 credited to my account the very next payment.
I have to say, in all my years of driving public vehicles Uber (with its hold on the rider's credit card detail) , have the best and quickest method to compensate drivers for these horrible incidents when they occur."
They can certainly have a video/image feed of the inside, to remote check if it is messy once the ride is over(can be automated to a good extent with image recognition), and retire it from service until its cleaned up.
In another case, a customer could actually throw garbage in the car and then report the car dirty - causing the previous customer to be fined. The only safety net against this I know is to record video inside the car while a customer is present.
you could have before and after photos taken for each trip.
We are a long way from where I would be willing to trust my life to self-driving cars - as a passenger, as another driver on the streets, as a cyclist, or as a pedestrian. Much farther away than these companies press releases make it seem.
Here's why. These driving algorithms are successful in large part because of data. They train their systems, such as visual recognition (what are the objects in the world around me), on millions of miles of visual data collected on the roads, most of it in California in the sunny daytime.
This means they are very likely to perform well in the average case when everything goes according to plan. And if deployed there they might live up to the hype and save thousands of lives compared to human drivers.
But now say you're in a major city in the midwest or northeast, for instance. It may be night time. It might be raining. There might be two feet of snow on the ground, narrow lanes, road signs covered up and unreadable. There may be a pedestrian crossing in dark colors. The street lines may be faded or nonexistent. There may be a street that is marked one way on the GPS map but is currently detoured the opposite direction due to construction.
There may be a policeman directing traffic. The police might pull the car over and direct it to a parking lot. There might be a fire truck or ambulance coming at an unusual time.
A computerized system trained on data can only perform well in situations very similar to its training set. But its vision will have a hard time recognizing objects it hasn't seen before. Its language processing will not understand unusual or novel road signs. Even if it recognizes the objects around it correclty, it lacks the "true" intelligence to deal with unforeseen situations falling significantly outside its training set.
I believe that cars are quite likely to run into novel situations they haven't experienced before, and I don't trust their reactions or decisionmaking in these scenarios. So I think what we have are self-driving cars that perform very well in the common, easy case, as we have already seen in numerous press releases, but are in my opinion very unpredictable in the long, fat tail of situations.
Police and emergency services will just coordinate with the "ATC" to pre-script routes differently depending on the situation.
In complex situations, flashing lights and honks are not enough, you need to verbally communicate with other drivers.
I would eat my hat if an AI could handle these kinds of situations.
So in short, I'll believe the hype when I see a video of a full auto drive through London at rush hour.
The Otto deal was set around incentives, it's not actually a flat 600 million dollar deal. If Levandowsky can deliver a Robotaxi OS that carries Uber to greatness, his networth could be astronomical. He's probably the top candidate in the world right now to lead an autonomous driving project to commercialization. He's got a vicious Randian streak and he's been at the bleeding edge of driverless vehicles since the Darpa days.
So, for example, chess AIs are better chess players than humans because they have longer lookahead, even though they're worse at analyzing a given board. The Watson AI didn't understand questions as well as a human, and had obvious comprehension failures such as the final jeopardy answer, but when its comprehension was good enough, it had a vast database that it could perfectly recall at very high speed, and those more than compensated for its comprehension problems.
Driving AIs are not as good at understanding what's happening around them, but they are constantly attentive in 360 degrees and have fast reactions, which (may, someday, but does not yet) compensates for their imperfect understanding of the world.
This AI is not generalizable to tons of other circumstances in which there is no obvious way to parlay the inhuman strengths of the AI into compensating for their weaknesses. As such, while there may be some other places where a driving-AI-like intelligence could be used profitably, there probably aren't many such places.
I am confused do they require a permit or is uber changing the claim/capabilities of the car to evade getting a permit?
"The company doesn’t require a permit from the California DMV to operate in the state, it says, because the cars don’t qualify as fully autonomous as defined by state law because of the always present onboard safety operator."
Uber could claim they need to record passengers in order to spot damage or dirtying of the cab. (Otherwise passengers could blame any damage on the previous occupant.)
This raises a good point i never thought about. If in some day all of our cars have self driving or self aware features, then that means nearly all locations in the city are within reach of cameras and possibly microphones. I'm not "a paranoid" about privacy stuff, but that is quite impressive nonetheless.
The future is going to be quite interesting. I always thought we'd end up with cameras on every street corner monitoring everything. I never thought our own cars could become every present monitoring devices.
You might find this article interesting: http://www.wired.co.uk/article/one-nation-under-cctv
> I heard there is a camera in the car. That’s correct. To learn more about how we can improve the self-driving experience for our riders, your trip may be recorded.
When picking up a ride they notify it's a self-driving Uber and let you cancel for free.
Even in highway driving you can lose traction in an instant if you hit a patch of black ice. Autonomous vehicles will need to be able to recover from a complete loss of traction safely. This isn't trivial - in fact it's probably the most complicated bit of driving I tend to do. Once you are sliding and your steering wheel becomes more of a suggestion than a command, the entire act of driving becomes a process of trying to coax the car off the road using a combination of steering, brakes, and even occasionally gas. I think it's possible for a computer to do this - but you can't avoid all slides just by driving slowly.
Then there's the plethora of other winter fun you run into with a vehicle: getting stuck (happens all the time on city streets) and all the techniques to get unstuck, going too slow and losing your momentum (and thus traction), having every indication you have traction and then discovering you actually don't (it's very easy to be driving at a "safe" speed and still slide through an intersection), white out conditions where you are guessing where the lane is... etc...
To be clear, I believe most of these conditions could eventually be handled by computers. I also believe a lot of people drive too fast in/on snow. However, winter driving is in no way simple. It's a problem domain unto itself, and one I've seen relatively little work being done on.
Winter driving is really no different from any other kind of driving in which the driver exceeds the limits of the vehicle's available traction. The methods of recovery are mostly well-known; the problem is more often the driver's inability to implement them in a timely manner.
Constant input from wheel speed and accelerometer sensor arrays, coupled with the vehicle's ability to individually brake/slow wheels (which also gives the vehicle the ability to accelerate individual wheels independently!) means that it could be a far easier 'problem' for self-driving cars to solve than it is for humans.
Again, if they're working on it :) But there's already been decades of work put into ABS, traction/stability control, etc.
It comes down to balancing, and we know we can build machines that can balance. Segways, bipedal robots, etc. So we have some of the tech and algorithms to do this in other applications, but applying it to autonomous vehicles will be its own beast.
This may seem like a straw-man argument, I don't intend it to be. I think self-driving cars will be on the whole better drivers, but these are also situations that I see as being extremely difficult for a computer to identify to the level a human driver is capable of.
You - and the other drivers on the road - have less control at all speeds and are always much closer to the limit of traction. Driving on winter roads is a lot like racing a car on a track with other drivers - everyone is near the limit of traction and a hazard can present itself very quickly. Having the right reaction at the right time helps, but planning ahead is more important. Daily driving in summer months is benign in comparison.
Google/Waymo and its successors are using pre-mapped courses with many heuristics and edge case tweaks. Routes are generated from existing resources (Google Maps, etc). Much effort is devoted to avoiding other vehicles and pedestrians. Much of the rules are based off of U.S. traffic rules, such as speed limits, stop signs, traffic signals, and lane markings.
They both share technology (computer vision, momentum/traction control), but I conjecture the bulk of the work for commercial autonomous driving was not related to the DARPA challenge and wasn't paid by its grants.
Sure, of course you're right that "the bulk of the work" is not related. But it's also no secret that the "sudden spike in interest in this technology" lucker referred to above happened because the US government paid for it to happen, as a means of advancing military vehicle automation technology.
Interesting. The US Military did spawn the self-driving Car Revolution with their DARPA Challange.
http://www.livescience.com/44272-darpa-self-driving-car-revo...
Does anyone have more info or speculation on the tech stack sitting in the trunk?
