home

Comment of the Week: How self-driving cars are actually going to work

Posted by on August 15th, 2014 at 2:39 pm

early-vehicle-lores
(Photo: Google)

Two big issues, gender in the bike world and the nature of Portland bike activism, generated lots of excellent perspectives from readers on the site this week. This one about the “thousand cuts” of being a woman was one of our most upvoted ever; this one early this morning about the recent history of Portland-centric bicycle advocacy groups is very persuasive.

But let’s finish the week on a lighter note, thanks to reader Jake.

Me: “car. Take me to IKEA”
Car: “location not found, please speak new command”
Me: “take me to IKEA…PLEASE. I need a new flimsy bookcase”
Car: “Confirmed. Route set for Fat Cobra”
Me: “What? No, cancel request. New request”
Car: “request cancelled, please state new request”
Me: “take me to IKEA near Airport Way”
Car: “processing request. Setting route for Airport”


Me: “No not the airport! IKEA! Jesus Christ…”
Car: “cancelling route, rerouting to The Church of Jesus Christ of Latter-day Saints”
Me: “NOOOOOOOOOOOOOOOOOOOOO!!!”
Car: “request cancelled”
Me: “Take me to the place with the cheap meatballs.”
Car: “setting route for IKEA”
Me: “Thank you. Finally….”
Car: “This vehicle has experienced a malfuncion and is no longer in service. To reset vehicle, open hood and unplug main power supply battery cable. Wait 30 seconds then plug back in and wait for rebooting process to complete. If problem sill exists, call the customer service phone number….which we wont tell you.”

Me: “Fu*k it. I am riding my bike.”

Enough said. Have a great weekend.


Email This Post Email This Post


Gravatars make better comments... Get yours here.
Please notify the publisher about offensive comments.
Comments
  • Hart Noecker August 15, 2014 at 2:47 pm

    The above is a humor reminder that tech can’t really function in the physical world for very long. Tom Vanderbilt discusses this a bit in ‘Traffic’ if I recall. Sure, a robot car with laser sensors can detect an object jumping out in front of it and stop – but it can’t tell the difference between a dog an old lady near the side of the street and use cultural information to make a prediction about the LIKELIHOOD of either jumping out into traffic.

    It’s the years upon years of cultural signals woven into our billions of neurons that allow us to make thousands of decisions every minute while operating a vehicle successfully. I wish I could say I’m excited to watch driverless cars fail, but seeing what the last hundreds of auto-induced damage has wrought, this just looks like out of the frying pan and into the other frying pan.

    Recommended Thumb up 1

    • Scott H August 15, 2014 at 3:18 pm

      Dude, no, please, just try and wrap your head around the fact that computer software is one of the solutions to our car-death epidemic.

      There were 33,561 motor vehicle deaths in 2012, 33,561. Humans are so inherently terrible at operating vehicles that google engineers will have an easy time making our roads safer. Inevitably, someone will eventually be killed by a google car, and there will be a media firestorm about that one death instead of the 100 others that day, I’m sure. But please, get your facts straight.

      Recommended Thumb up 27

      • Ian August 15, 2014 at 4:47 pm

        I don’t see any facts in your story, just suppositions. There is nothing inevitable about the ability of machines to be better than humans at driving a road vehicle.

        Recommended Thumb up 4

        • Scott H August 15, 2014 at 5:38 pm

          The fact that you’re communicating with me on this very web site, instead of mailing me a letter, is fact enough. Humans have been using technology to solve problems forever. So yes, I’m supposing that trend will continue.

          Recommended Thumb up 15

          • Hart Noecker August 15, 2014 at 7:33 pm

            That’s probably the most absurd non-sequitur I’ve heard all week.

            Recommended Thumb up 4

          • 9watts August 16, 2014 at 10:17 am

            “Humans have been using technology to solve problems forever.”

            Sure, but where are you drawing the boundaries of that inquiry? How are you measuring the beneficial? detrimental? effects of this technological progress?
            Warfare = (rocks -> machine guns -> carpet bombs)
            Transportation = (feet -> bike -> train -> auto)
            Agriculture = (hoe -> horse -> tractor -> herbicides -> GMOs)
            Medicine = (herbs -> baby formula -> antibiotics -> Prozac)

            There is nothing inevitable about technical change, nothing inherently beneficial. Lots of changes follow predictable paths, but the idea implicit in your post that these changes automatically correspond with some unspecified improvement is without basis.

            While I like blogs as much as the next person, there is nothing self-evidently ‘better’ about a blog than a letter. They certainly differ along many axes, but I don’t agree that we’re seeing any neat progression toward some ever-more-refined future state.

            Recommended Thumb up 5

            • Scott H August 18, 2014 at 12:00 pm

              You’re smart enough to grasp my point, that we will soon have the opportunity to save thousands of lives a year, which means you must be arguing with me just to argue with me.

              Recommended Thumb up 4

              • 9watts August 18, 2014 at 12:23 pm

                I don’t think so. I am not persuaded that self-driving cars will ‘save thousands of lives a year.’ Just as I am not persuaded that the Green Revolution (or, more recently, GMOs) will save millions of people from starvation. Or that nuclear power will make electricity ‘too cheap to meter.’ Or that electric cars will avert climate change. We’ve been promised these kinds of things by capitalists and their government boosters for generations. This sort of thing almost invariably turns out differently than we are led to believe they will.

                Recommended Thumb up 5

                • Scott H August 18, 2014 at 4:18 pm

                  I’m not talking about climate change, or world hunger, or an infinite source of energy. It’s already common knowledge that computer software can perform tasks, like tracking objects, and avoiding collisions, with more precision, reliability, and speed than a human. The technology will eventually be perfected. Anyone that doesn’t understand this likely doesn’t understand computer software.

                  Recommended Thumb up 0

                • 9watts August 19, 2014 at 12:45 pm

                  “Anyone that doesn’t understand this likely doesn’t understand computer software.”

                  I admit readily that I don’t understand computer software. But I will also submit that the problem before us isn’t, chiefly, a software problem.

                  These computer cars, after all, are going to have to interact with a whole lot of hardware: air bag sensors, brake fluids, children playing in the street, blind people, people riding bicycles, other malfunctioning computer cars, downpours, electrical storms, potholes, improperly entered code, heat, hackers, tire blowouts, icy driveways, ABS sensor glitches…..

                  I am amused that you seem to think that all we need to know about this is that software is capable of ‘handling this.’

                  Recommended Thumb up 0

                • Scott H August 21, 2014 at 4:47 pm

                  Oh jeez, look what you did, you used up all of the indents. Now I can’t even tell if I’m agreeing or disagreeing with you.

                  Recommended Thumb up 1

                • 9watts August 21, 2014 at 4:49 pm

                  Let’s call it a draw -

                  Recommended Thumb up 1

      • Hart Noecker August 31, 2014 at 1:15 pm

        All this blind faith technology regarding driverless cars is pretty amusing when you consider the fact they can’t even drive in the rain.

        Oops.

        http://www.autoworldnews.com/articles/8817/20140830/googles-self-driving-car-cant-navigate-heavy-rain-or-most-roads.htm

        Recommended Thumb up 1

    • Anne Hawley August 15, 2014 at 4:02 pm

      Yet. The key word is always “yet.” Tech can’t really [insert function here] YET. It – whatever “it” is, whether self-driving cars or medical diagnostics on your iPhone – doesn’t have to be perfect to replace a human being; it just has to be better than human. In the case of cars and driving, that’s a pretty damned low bar.

      Recommended Thumb up 15

      • q`Tzal August 15, 2014 at 4:23 pm

        And then the automobile insurance companies have another reason to jack up your rates:
        “The numbers here show that you have a higher risk factor than the autonomous driving module we’ve offered to purchase and install in your car free of charge. Our records show you drive too fast in inclement weather and heavy traffic. Also there is the matter of the speeding ticket and public videos of your parallel parking.
        “On your current rate plan the autonomous driving module is not optional. If you prefer we can upgrade you to our premium driver coverage but this requires a 2 week driver education course and yearly in class refresher coursework. Of course it charges by mile driven manually and we reserve the right to cancel your policy immediately upon conviction of any traffic infraction.”

        THIS is how the self driving car “thing” is going to go.

        Recommended Thumb up 10

    • Paul Atkinson August 15, 2014 at 4:21 pm

      Specifically with regard to your example (a dog vs. a human)…yes, in fact, it can tell the difference and prepare correctly.

      There may be things that are difficult to teach a computer but that’s not one.

      Recommended Thumb up 12

      • Karl Dickman August 16, 2014 at 9:38 am

        Thanks for bringing this up, Paul. Dog vs. grandma isn’t a “yet” thing, it’s a “now” thing.
        And Hart: my take has always been that driverless cars may not necessarily be better at driving than a skilled, sober, 100% focused driver, but better than the “average” driver is a much lower bar to clear.

        Recommended Thumb up 4

  • Dan August 15, 2014 at 3:13 pm

    I assume every moving object is likely to jump out of in front of me without looking.

    Recommended Thumb up 2

  • John Liu August 15, 2014 at 3:27 pm

    Google’s prototype autonomous cars have self-driven over 700,000 miles with only one accident, in which the car was rear-ended at a light. There has been one accident when one of the cars was being manually driven.

    Recommended Thumb up 15

    • Ian August 15, 2014 at 4:50 pm

      Really all this speaks to is how careful Google have been in testing their technology, which is admirable but not indicative of real world conditions in my opinion.

      Most of those were highway miles (much easier to program for) and the extensively mapped area around Google’s South Bay campus. The amount of very specific data and input that it takes to enable a machine to drive a car in urban situations is still a way off.

      Recommended Thumb up 2

      • John Lascurettes August 15, 2014 at 5:28 pm

        Ding ding! We have a winner. Google has already admitted that their tests are only near-campus specific. They don’t have enough ubiquitous data to open this up. Also, do you think we really should let Google be in control of the data set on this – if we had autonomous driving ever be a compulsory function of automobiles, then it would have to be a government-owned or government-regulated commons to ever have it work in the public’s best interest – yes, the government is good for some things and this is exactly what it’s designed for.

        Recommended Thumb up 1

        • Pete August 15, 2014 at 10:25 pm

          I don’t agree that government should ‘own’ this data or technology. Regulate it, yes, but modern government regulations rely VERY heavily upon the expertise of technologists in private industry. Most regulatory guidance, be it power & water, oil & gas, medical, aerospace, industrial cybersecurity, etc., is created and managed by technologists in private industry. It is mostly funded and administered by government entities (like NIST, DHS, FDA, FAA, NRC, FERC, etc.), with the exception of (sometimes in cooperation with) national labs, NASA, etc. If you want an example of government ‘owning’ technology take a look at the Obamacare rollout, and then read up on the companies from which they hired ‘experts’ to fix the problems (hint – one of them makes autonomous cars).

          What scares me more on the roads is the number of times I’ve seen Google’s shopping van drivers careening around the south bay, knowing full well that vehicle telematics to monitor driver behavior – and the data analysis tools behind them – are already in wide use among delivery companies, garbage collectors, short-haul truckers, etc. The last time I saw one of their prototypes I was doing 65 MPH up I-85 N and it flew by me like I was sitting still. I do agree they still have a long way to go with T&D before being allowed to coexist with us on the road, but I’m way more afraid of their drivers than I am of their driverless cars.

          Recommended Thumb up 5

  • q`Tzal August 15, 2014 at 3:55 pm

    The possibility of error is no reason to settle for the certainty of error.

    Recommended Thumb up 10

  • peejay August 15, 2014 at 4:09 pm

    Interesting to note that the first crash caused by a self-driving car will almost certainly create a storm of bad press, but no bad press happens during the many frequent crashes caused by human-driven cars. I’ve always thought that if cars were introduced today, the quickly mounting death toll would take them off the roads within weeks.

    Recommended Thumb up 16

  • Jack August 15, 2014 at 5:13 pm

    Hart Noecker
    The above is a humor reminder that tech can’t really function in the physical world for very long.

    This assertion is proven entirely false by so much existing technology that does function in the physical world all day, every day. You need to always keep in mind that the things we commonly refer to as technology are always in flux. The radio is technology, but it’s probably not what you were referring to. High-tech becomes low-tech when the newest tech comes around.

    Recommended Thumb up 16

    • Hart Noecker August 15, 2014 at 7:37 pm

      Sorry, I was referring to technology operating on its own without constant monitoring and adjusting by people in the real world – as a driverless car would have to be. It’s a nice dream, but as I said above, there’s no computer that can do more than react in real time, which obviously isn’t enough to safely operate a 1 ton vehicle moving at speed.

      Recommended Thumb up 1

      • Jack August 15, 2014 at 8:01 pm

        Again, where are you getting the idea that predictive software doesn’t, and can’t exist? It does, and therefore, can.

        Recommended Thumb up 13

        • Karl Dickman August 16, 2014 at 9:57 am

          Here are some predictions that I have had to make. Every single one of these predictions could be made by a computer. Google’s car can probably already deal with 1-4 (as well as grandma/dog) already:
          1. Approach to the Hawthorne Bridge: there might be a walker or biker hiding behind the Marquam columns.
          2. Water Ave. & Yamhill: Drivers are just getting off the freeway so they usually blow through the stop sign.
          3. The people on the sidewalk are only 10 ft from the intersection, so they might make a hard left turn into the crosswalk because they can’t hear me coming.
          4. That biker with the earbuds might jump into my lane without warning to make a turn.
          5. A doe just crossed the road. There’s probably a few more deer with her.
          It takes a huge investment of money and time to make this work, but that’s not the same thing as impossible, especially with Google and Tesla and the Feds clearly willing to fork over a lot of time and money.

          Recommended Thumb up 4

          • wsbob August 16, 2014 at 8:51 pm

            Predictability of autonomous vehicles may be great for people that don’t like following the rules of the road. Autonomous vehicles supposedly are being designed to stop for things like kids and dogs darting out in front of the vehicle.

            So, someone in a non-autonomous vehicle feels like cutting off the robot car? I suppose they can do that knowing, the likelihood it will stop to avoid a collision, is greater than if a living human were driving the car.

            Recommended Thumb up 1

            • Spiffy August 18, 2014 at 2:24 pm

              unless I’m the human… my camera is rolling… cut me off all you want, but I have you on video when you slam on the brakes and I hit you…

              Recommended Thumb up 1

              • wsbob August 19, 2014 at 10:48 am

                “…but I have you on video when you slam on the brakes and I hit you…” Spiffy

                The ‘beauty’ and fallibility of autonomous car technology, is that vehicles so equipped, won’t allow rear end collisions to occur.

                They may by design, be traveling so close together that there won’t be enough space between cars for a car in an adjoining lane to abruptly cut in, or they’ll respond to a cut in by having long strings of vehicles simultaneously begin braking to safely accommodate the action. Autonomous car technology could have the commute become a car train.

                Recommended Thumb up 0

                • 9watts August 19, 2014 at 11:14 am

                  “vehicles so equipped, won’t allow rear end collisions to occur. ”

                  You folks are hilarious in your credulity. You really think that these imaginary vehicles are *never* going to have brake failures? transmission oil leaks? electrical malfunctions? (computer glitches?) Can you imagine overlaying these everyday occurrences of today’s mundane cars (or computers) onto something as mind-bogglingly complex as what Google is fooling around with? How long, for how many miles, do you expect these vehicles to operate? What happens when there is an electrical trouble? Who notices and takes care of it? Does the car drive itself to the shop? Who performs the repair? Right now mechanics with 100,000 manufacturer-certified diagnostic computers can’t figure out why the check engine light stays on in my mom’s VW TDI. This uncertainty has lasted for almost fifteen years*. This is all so absurd and ridiculous and fanciful.

                  *Which means it is an awfully reliable car despite the failure of the check engine light system, but it also gives me pause when we’re talking about autonomous vehicle technology.

                  Recommended Thumb up 0

                • dr2chase August 19, 2014 at 3:32 pm

                  @9watts – if we’re down to only worrying about mechanical failures, we’re doing very well, even ignoring the fact that a computer will do a better job catching those early than the average human. How often does a car’s brake failure occur in the brakes, versus their application? Today I saw a guy from a side-street stop just miss a guy on a Hubway bike — six inches more before stopping and he would have hit him.

                  As far as detecting failure before it happens, again, a low opinion of the average driver is justified. I worked on cars when I was younger, I hear diagnosable problems on the road every day, no instruments required. The two most common are worn brake pads (squee-squee-squee-squee), or a constant-velocity joint that’s lost its seal and all its lubrication (thunk-a-duh-thunk-a-duh-thunk-a-duh as the car turns under power).

                  Recommended Thumb up 1

                • wsbob August 19, 2014 at 4:01 pm

                  “…How long, for how many miles, do you expect these vehicles to operate? What happens when there is an electrical trouble? Who notices and takes care of it? Does the car drive itself to the shop? Who performs the repair? Right now mechanics with 100,000 manufacturer-certified diagnostic computers can’t figure out why the check engine light stays on in my mom’s VW TDI. This uncertainty has lasted for almost fifteen years*. …” 9watts

                  On board sensors could be made to monitor cars’ condition to operate safely on the road, and regulate their access to the road. If an autonomous cars component malfunctions en-route, and isn’t in sufficient condition to allow the car to operate safely, the on board systems could bring the car to the side of the road or a safe emergency location. It may have to be towed to a repair shop in the event of a component failure that could critically affect safety.

                  Mechanics’ diagnostic tools get better and better, as does vehicle operational systems and components. At 15 yrs old, your mom’s TDI’s control and monitoring systems are probably antiquated and essentially obsolete in comparison to current electronic vehicle systems. Similar to pc’s and laptops that new, six or seven years ago, were the hot item, but today, are obsolete.

                  Recommended Thumb up 0

                • 9watts August 20, 2014 at 6:47 am

                  ” At 15 yrs old, your mom’s TDI’s control and monitoring systems are probably antiquated and essentially obsolete in comparison to current electronic vehicle systems. Similar to pc’s and laptops that new, six or seven years ago, were the hot item, but today, are obsolete.”

                  This is what I’m getting at. But, see, the check engine light snafu started at 75,000 miles, which was about four years after she bought the car. She’s at 245,000 miles now. The entire time the light has stayed on, and the mechanics eventually gave up trying to find or fix the cause of this. Mechanical systems can be maintained, repaired pretty much indefinitely, and with tools that are reasonably affordable to an individual (mechanic or not). Once we started adding computers into these otherwise primarily mechanical systems the ability to troubleshoot and repair took a turn in a completely new direction. Now we need diagnostic computers, apparently, and when they don’t work, everyone just shrugs. The self-driving car, to me, appears to take another leap in this direction. I am well aware that there are many people who think this is a salutary thing; I’m just registering a contrary opinion.

                  Recommended Thumb up 0

                • dr2chase August 20, 2014 at 8:45 am

                  @9watts – I’m not a fan of what I’m about to do, which is say “but we’ll do this different better thing”, but we’ll do this different better thing. Infernal combustion machines are a nightmare of complexity and things that can go wrong (*), and it is frankly miraculous that we have gotten them to work as well as they do. Almost certainly the self-driving cars will be electric, and that will eliminate many annoying problems at one fell swoop, and simplify lots of control and instrumentation (for example, measuring power in is easier, just monitor voltage and current, much less need for mechanical/chemical sensors).

                  (*) My personal favorite, happened to me driving 60mph on 101 in Silicon Valley, is to have your fuel line come loose downstream of the fuel pump. Instant loss of power and the car is of course filled with gasoline fumes. I steered out of traffic, pulled over, found the loose fitting, scratched it with some pliers to make it grip better, shoved it back in, waited for the gasoline to evaporate, resumed my trip, and that fitting never gave me another problem. I think your point is that a robot might not deal with that correctly (I am still somewhat proud and surprised that I did), and my point is that it we should eliminate the problem entirely by going electric.

                  Recommended Thumb up 0

                • wsbob August 20, 2014 at 10:42 pm

                  9watts@http://bikeportland.org/2014/08/15/comment-week-self-driving-cars-actually-going-work-110106#comment-5396373

                  Same explanation applies to the old TDI’s long standing problem. 15 year old electronic design in motor vehicles is ancient. I don’t know it for a fact, but I imagine more recent car’s systems are much more accurately and efficiently diagnosed. I understand my own, almost 20 yr old pickups’ system design is just one year short of a manufacturer transition to one that allows a mechanic to hook up a single instrument to the vehicle and diagnose a huge range of problems.

                  How good a human development it is, I’m not sure, but I’m very confident related industries and government together, will be able to have autonomous vehicles become the vehicle of choice, and obligation for many people. Such vehicles will be able to be more reliably fixed, but only to a point, because the industry would not like giving up built in obsolescence.

                  Recommended Thumb up 0

          • Hart Noecker August 18, 2014 at 5:55 pm

            That’s science for ya, always ‘could be’ and never ‘should be’.

            Recommended Thumb up 0

  • Kyle August 15, 2014 at 9:30 pm

    I think driverless cars are inevitable. All you have to do is look at drones. How many drones accidents do we hear about? I know they are piloted remotely but how different is that from having a human sitting in the driverless car? Couldn’t we just override the auto drive as needed? Maybe creating a reliable selfdriving car is more is complicated because of all the potential random external enoucters, but the fact that Google has driven their cars 700,000 miles with out causing an accident indicates they are coming.

    Recommended Thumb up 1

  • Matthew Rogers August 15, 2014 at 10:35 pm

    Autonomous vehicles will come, it’s just a matter of time. I enjoy all modes of surface transportation for people and freight. I prefer biking for flexibility, health, and cost savings for both me and the planet, but cars and trucks have their places too.

    Predictive technologies are already in use for optimizing certain systems based on upcoming terrain in vehicles already: cruise control speed in semi-trucks (Freightliner’s RunSmart Predictive Cruise [disclaimer: I work for Daimler Trucks, but this is comment is my own opinion]); BMW I believe is working on a transmission that predictively shifts depending on upcoming curves and driving style.

    There’s also a lot of work being done on pilot studies for vehicle-to-vehicle and vehicle-to-infrastructure communication that allows vehicles to alert drivers or even react automatically to hazards that humans couldn’t see. Ann Arbor has a fair number of vehicles and intersections wired up for V2V and V2I technology. It allows for alerts to be given for cross-traffic right-of-way violations, alerts for braking vehicles multiple cars ahead (imagine seeing brake lights *through* a car or two ahead). It also allows for more “mundane” alerts, like blind spot warning and curve speed advisories for example.

    Interestingly, Daimler Trucks, though the Mercedes Benz truck brand just released some videos and details of their FutureTruck 2025 concept, which includes autonomous highway driving, incorporating a ton of radar and camera sensors, as well as V2X technology: https://www.youtube.com/watch?v=WxzVLFoav8M

    Recommended Thumb up 7

  • Matthew Rogers August 15, 2014 at 10:43 pm

    Speaking more specifically about Google’s autonomous vehicle program, this video really shows how much information is being recorded, identified and processed. The amount of vehicles, signage and road conditions that it has to store at any given time is staggeringly large and it’s neat to see them all identified and tracked in the computer view. The visualization of the fencing is pretty neato, as well.

    It looks like they’re able to do a pretty good job of predicting bicycle movement based on hand signals, too.

    https://www.youtube.com/watch?v=csvt6JBAwBk

    Recommended Thumb up 3

  • dr2chase August 16, 2014 at 7:54 am

    As noted above, Google makes this work in their test area with enormously detailed maps of Mountain View. I think the trick for bootstrapping beyond Mountain View is to sell to people with horrible commutes (and these cars will still have conventional driver controls), and the drivers will “train” their cars by driving their horrible commute a dozen times while the car collects data and sets it back to the mother ship.

    And then that route is known, and that customer is happy.

    Repeat for sufficiently many customers, assume they will fill in ever more data as they drive other random routes, and in a few years all economically interesting parts of the US will be mapped.

    Recommended Thumb up 2

    • Spiffy August 19, 2014 at 8:08 am

      Google street view is already fairly detailed maps of every street…

      Recommended Thumb up 0

      • dr2chase August 19, 2014 at 1:05 pm

        More detailed than Street View: http://www.theatlantic.com/technology/archive/2014/05/all-the-world-a-track-the-trick-that-makes-googles-self-driving-cars-work/370871/

        What I did hear (last week, from someone with a professional interest in this who does not work for Google) is that one peculiar problem is snow storms. Not so much snow on the ground, but falling snow, because the optical sensors don’t deal with it so well (falling snow is also incompatible with a helmet spotlight at night; the brightly lit snowflakes right in front of your face make it impossible to see anything).

        Recommended Thumb up 1

        • 9watts August 19, 2014 at 1:13 pm

          “We tell it how high the traffic signals are off the ground, the exact position of the curbs, so the car knows where not to drive,” he said. “We’d also include information that you can’t even see like implied speed limits.”

          Except that a few months later someone from PBOT comes along and installs new traffic signals, or puts in a bioswale where the curb used to be. Or the traffic signals stop working (happens quite often, actually), or the stop sign at a four-way stop is obscured by bushes when the google car drove by….

          All of this is coded (it sounds like) by humans (= lots of jobs), but they get tired, make errors.

          Recommended Thumb up 0

          • dr2chase August 19, 2014 at 1:27 pm

            I am almost certain that this is all automated, and that the Google cars will be in constant communication with the mother ship to keep the databases updated on changes to the road. This is not that much different in principle from how they keep their search index updated as websites change more or less gradually over time.

            Recommended Thumb up 1

            • 9watts August 19, 2014 at 1:31 pm

              I’m sure you’re right. I just think we’re forgetting the fact that cars—quite apart from any computers that may get incorporated, and which have their own reliability issues—break down, perform unexpectedly (Cartalk, anyone?), and all this euphoric buzz about how they will be able to handle all of this leaves me unimpressed.

              Recommended Thumb up 0

              • dr2chase August 19, 2014 at 2:05 pm

                I’m somewhat more optimistic about relative success only because I have such a low opinion of human drivers, and the videos I’ve seen (in highly-mapped Mountain View) seem to show the robo-car doing some completely intelligent things around humans (including bicycles).

                I think people are too quick to assume that the goal for robot cars is to drive exactly like humans. That’s clearly wrong, because humans kill a lot of people on the roads. I expect robot cars will drive differently in two ways. First, all the avoidable dumb errors that humans make, robot cars won’t make. No tailgating (of humans), speeding, rolling stops, bad passes, poor lane discipline, distracted driving, sleepy driving, or drunk driving. Second, they will probably take an algorithmic approach to crash avoidance. If you look at how drivers behave (especially in the US) there is this assumption that everything will go swimmingly, and that all rules (both real and imaginary) will be followed. And when things go wrong, “I couldn’t stop in time” (which I think is garbage, but that’s how we drive). A robot car can take a different approach of cataloging all the things that might go wrong in the next second, and modify speed/path to ensure that stopping before disaster is always possible. This probably results in slower travel in urban areas.

                Thinking about this problem (for a while now) has changed the way I bike. I can’t reliably track 3 randomly moving things at full speed — so if I see 3 or more children or dogs on the MUP ahead of me, I slow way down.

                Recommended Thumb up 3

                • wsbob August 19, 2014 at 3:03 pm

                  “I think people are too quick to assume that the goal for robot cars is to drive exactly like humans. That’s clearly wrong, …” dr2chase

                  Autonomous cars will be operated the way computers operate. Very logically and efficient. Goodbye to freedom of the road. As I said earlier people traveling by non autonomous vehicle, may be able to exploit the limitations of autonomous vehicles around them to their advantage or pleasure, or maybe not. References to Asimov’s three laws of robotics, though science fiction, may not be limited to being purely a joke as the age of autonomous vehicles approaches.

                  Recommended Thumb up 0

  • Panda August 16, 2014 at 11:46 pm

    Commercial flights already take off, fly and land autonomously. Not all, and all the planes have pilots, but many do

    Recommended Thumb up 2

    • 9watts August 16, 2014 at 11:59 pm

      You guys are killing me. Are you really trying to make the case that airplanes are a reasonable analog to cars when it comes to navigation?

      Did you forget the FAA? Control towers? The third dimension? What is called bird ingestion in the jargon?
      http://en.wikipedia.org/wiki/Bird_strike

      Recommended Thumb up 2

  • CJ August 17, 2014 at 1:38 pm

    All that I know is that my computer crashes a lot more often than my car or bike.

    Recommended Thumb up 0

    • Pete August 18, 2014 at 11:38 pm

      Your computer’s software was designed to be affordable, not safe.

      Recommended Thumb up 1

    • GlowBoy August 19, 2014 at 10:18 am

      You must be running Windows. My MacBook is 7 years old, and has crashed about the same number of times as my bike in that time. At work I regularly log into Unix systems that have been in continuous operation for years without a crash.

      Recommended Thumb up 1

  • GlowBoy August 18, 2014 at 10:12 am

    I think driverless cars are coming, but that full adoption will take longer than proponents hope. It will probably be at least a decade before the technology is foolproof, and then what?
    - Historically, new safety improvements (ABS, airbags, stability control, etc.), have initially appeared on luxury cars, taken 10-15 years to trickle down to less-expensive cars.
    - Then it takes a few more years before the technology becomes mandatory on all new vehicles.
    - At the moment, the average car on the American road has crept beyond 10 years old, despite Cash to Clunkers having taken several million older vehicles out of service. So allow another 10 years before there are more autonomous cars than not. Now we’re approaching half a century out.

    Don’t get me wrong: I do think I will live to see the day when autonomous cars are commonplace.

    But I doubt I will live to see the day when they are universal.

    And I am sure I will not live to see the day when (to the point of the story that started this discussion) the vast majority of people see their cars so solely in utilitarian terms that they stop owning them, and only use short-term rentals. 95% of vehicles in service at any given time?!? That may prove the case in a few pockets of dense urbanism, but won’t be the norm across America.

    If anyone here has plans to time-travel back to the 1960s or 70s, please tell the futurists I’m still waiting for my monorail.

    Recommended Thumb up 0

  • Sam Churchill August 18, 2014 at 2:35 pm

    Here’s my proposal for an autonomous car test on Hayden Island. http://i5bridgecam.wordpress.com/selfdrivingcars

    One advantage of battery-powered, rubber-wheeled vehicles is that an expensive I-5 bridge is not required. No overhead line for power. No tracks.

    They could offer safer, faster, cheaper connection to public transit while providing improved bike access.

    Recommended Thumb up 2

  • Jake August 18, 2014 at 4:07 pm

    Wow. my quote made front page blog!
    Thanks Michael.
    I was going to type a serious in-depth analysis about cars and people bikes and autonomous this and google that…blah blah… But looks like we have enough of those party killers on this blog. C’mon folks, your internet comments arent going to make a hill of beans difference in how the future shakes out, so RIDE YOUR BIKE and have a fun!

    Recommended Thumb up 4

    • Pete August 18, 2014 at 11:40 pm

      “your internet comments aren’t going to make a hill of beans difference in how the future shakes out”

      Google just heard you say that… and Google knows where you live. ;-)

      Recommended Thumb up 1

  • Ryan Francesconi August 19, 2014 at 11:10 am

    We’ve had this all along. It’s called public transportation. Unfortunately in this country it sucks. Self driving cars aren’t a solution to anything long term. They sound more like a solution to the problem of texting while driving. Now you can be a zombie even when driving to IKEA. No wonder google wants to develop it.

    The reason that there are so many deaths in motorized vehicles is that the users don’t earn their right to use them. It’s too easy to be licensed. Auto industry lobbyists (and gun lobbyists) will ensure this death rate remains high.

    One thing that is brilliant and correct about the bike is that you must earn your speed and handling through years of hard work.

    Recommended Thumb up 0

    • Dan August 19, 2014 at 3:23 pm

      or get an E-bike and haul ass right off the bat.

      Recommended Thumb up 0

  • Gumby August 20, 2014 at 12:35 pm

    Two steps forward, one step back. There are always unintended consequences with the solutions to any problem whether it’s new technology or not(except bicycles of course). I saw a video segment where Jay Leno explained how much of a problem horses were in New York city before the advent of cars. They left a mess everywhere and when one died, it was often just left where it fell. Cars were the saviour for all that. Automated cars will undoubtedly solve a lot of problems and invent brand new ones for us to solve – and so the story goes.

    Recommended Thumb up 0

  • wsbob August 21, 2014 at 10:46 pm

    In the NYTimes today, a story about future cars’ electronic systems communicating with each other towards achieving a higher level of personal and road safety. Link:

    http://www.nytimes.com/2014/08/21/business/new-era-in-safety-when-cars-talk-to-one-another.html

    Just the kind of thing among others, that’s been speculated on in earlier comments to this bikeportland story.

    Recommended Thumb up 0

- Daily bike news since 2005 -
BikePortland.org is a production of
PedalTown Media Inc.
321 SW 4th Ave, Ste. 401
Portland, OR 97204

Powered by WordPress. Theme by Clemens Orth.
Subscribe to RSS feed


Original images and content owned by Pedaltown Media, Inc. - Not to be used without permission.