[Geohot] Selling His “Self-Driving” Car Tech For $1k By New Year

This week [Geohot] announced the launch of his self-driving car hardware. This is the natural extension of his proof-of-concept shown off in December which he parlayed into a Silicon Valley startup called comma.ai. [Geohot], whose real name is [George Hotz], is well known for jailbreaking the iPhone and making Sony look like idiots when they retroactively crippled Linux support on PS3. He has hardware chops.

Initial self-driving add-on hardware only works with Honda and Acura models that already have lane-keeping assist features because those vehicles already have built-in front radar. The package, which replaces the rear view mirror, adds a front facing camera. Those lucky (or brave, foolish, daring?) beta users can trade $999 and $24/month for what is currently a green 3D printed enclosure with some smartphone-like hardware inserted.

The company has taken an interesting approach to acquiring data needed for this particular flavor of self-driving. [Hotz] is teasing a chance at beta test invites to those who contribute driving data to the company. This is as simple as downloading an app to your phone and letting it roll from your windshield as you go bumper to bumper from Mountain View to San Francisco. That’s right, the plan is to support just that stretch of the nation’s highway system — although [Hotz] did make a brazen estimate of 90% of commutes for 90% of users within a year. Hey, it’s a startup so it’s either that, selling to a bigger fish, or closing their doors.

That narrow route support is actually an interesting constraint. In fact, the company is most interesting because of its chosen constraints: a small subset of cars, a chosen stretch of highway, and dare we say sanity when it comes to self-driving expectations. Grandiose claims have the general public thinking a vehicle with no human driver will slide up to your stoop and take you anywhere you want to go. That is a dauntingly difficult engineering challenge (dare we say impossible). What [Hotz] is selling is a more stress-free commute, not a nap in the back seat. You still need to be paying attention at all times.

Will this system work? Undoubtedly the engineering is possible (Tesla is already doing it). The biggest question mark that remains is human nature. This system demands your attention even though you’re doing nothing. That seems unrealistic — users are bound to lapse in attention much more frequently than if they were the primary driver. The question then becomes, will people pay attention at the very rare yet very crucial moments, and can a system like this prevent more fatal accidents than it causes?

[via Engadget]

58 thoughts on “[Geohot] Selling His “Self-Driving” Car Tech For $1k By New Year

  1. why do people buy cars if they don’t want to drive them?
    how about an app that lets these people all club together and rent a coach or something instead of buying cars they hate driving so much. maybe they could start a rail company or something.
    or is it just the idea of robot cars that people jerk themselves silly over?

    1. I think it is mostly for highway driving; it gets boring quickly.
      Or traffic jams that are stop-and-go just put people in a bad mood to have to deal with. Or older people/disabled will really benefit from self driving autos.

      That said, I wouldn’t trust a computer to slam on the brakes or take the curb when it will matter. One glitch and the system goes ‘HELP! I DON’T KNOW WHAT TO DO! I SEE OHH NOO A CARDBOARD BOX AAAHHHHH!!
      Vehicle slams on the brakes and someone else slams into the ‘smart’ car.

      1. I can think of a few “near death experiences” that I would not trust current tech in self driving cars to handle.

        Driving along in the city, windy, see a 8×4 sheet of plywood peel off the 4th story of some construction, come guillotining down to where I would have been in half a second if I hadn’t braked. (Probably a good thing for me it bounced too)

        Caught in a severe thunderstorm, lightning hit the power lines behind me, transformer on a pole blew up just ahead of me and a wire came snaking down, just managed to swerve around the end of it. Same storm seconds later, severe downbursts start slamming and bending the trees immediately ahead, big trees doing the wacky waving inflatable arm flailing tube man thing, and one has a major branch over the road and I don’t like the look of how much it’s whipping about, mashed the gas this time, crashes down right behind me.

        ABS, loose gravel into hairpin on steep downhill, ze brakes, zey do nuzzing, handbraked it round.

        Then there’s those kind of chessplaying type decisions, where you basically see the 10 vehicles ahead of you, know how they have been driving, and either move lanes or stay put, because some situation up the road is going to cause someone to be where you are now, or in the space you wanted, but decided to hang back. So you do, and yes, situation develops where you’d have been hard on brakes and cursing, taking the ditch as the soft option, or tapping fenders, if you had done the other.

        Oh, another, super, highly, glassy slick ice at an intersection. I saw it, I very delicately brought the car to a stop about 10 feet early. Sitting there waiting for the light and the damn wind starts me sliding again, into traffic, slight slope, manage to shimmy it over into the curb to stop me.

        Yeah, also, “knowing when not to stop” I’ve had to choose to run reds before because I didn’t think the idiot right on my ass would manage to stop, one time I did and 3 cars followed.

        1. My biggest concern is weather!
          Hail, heavy rain, lightning strike (flash might blind the camera), those days when tumbleweeds are blowing across the road.
          The system needs to have a full-override option for me to even temporarily trust the automated system; if we’re talking about a decade from now.

          P.S.: I hate ‘black ice’. Almost smashed a pole because I shifted into the wrong gear and completely lost traction. *spins*
          Automatic traction control is pretty awesome on newer vehicles though.

        2. Pondering that last, also makes me wonder how willing autonomous systems would be to break the letter of the law to keep occupants safe. Thinking there’s been occasions where I have had to cross double lines or been going the speed limityet had to mash the gas suddenly.

          One specific time I had to “teleport” into a restricted access commuter lane when someone tried to merge into me. Now the maneuver I mentally refer to as the teleport, I’ve had to instinctually resort to about 3 times…. it’s kind of slam the wheel half lock and back again very quickly. I am unsure whether they would fit steering actuators capable of it. This would be at about 70mph… You could describe it as “try to flip the car but just before the tires come off one side of the rim try to make them come off the other”…. So yeah that’s not going to be within any parameters the autopilot is going to be allowed to use. I have a hard time analyzing exactly what I do, there is some feeling of the pendulum mass of the car and the compression of the suspension, it’s kind of like you semi whip it and leap it rather than steer in a conventional sense. Would be a bit of an outside the envelope edge case to program. Last occassion it happened was in a vehicle with stability control and I was hugely gratified it let me do it. I actually think it beat it, the point where it would have realised the tire slip was the point where I was whipping it back the other way again.

          I also think it’s near impossible to practice without that nerve jolt of survival reflexes, you’re almost bound to do it too slow. Rally drivers might manage it “on demand” as it were.

        3. i would guess that there are self driving cars on the roads today somewhere that have experienced similar things and gotten through them, not to mention all the cases where the human reaction time is just too slow from a raw time to action perspective.

          i don’t think it will be that long before self driving cars are better than 90% or more of drivers in virtually every situation , hell at this point there is no board game on the planet that ai cant beat humans in consistently, their pattern recognition is also evolving and improving very fast.

    2. – self driving technology might open up to a market of people that have large retirement saving – people too old and too unsafe to drive
      – trucking fleet
      – car sharing/rental/taxi etc. Get rid of the driver, then you save a lot of salary

    3. Driving is a menial task.

      The claim “i like driving” is basically a meme created by people’s orthodox Luddite fear of the unknown.

      Self driving cars do come with a host of concerns though – privacy, safety, etc.

          1. I learned on manual and was driving manual vehicles for 15 years.. now only occasionally. Automatic is so much more relaxing and cruise control even more so. If that transportation device was able to steer itself on top of that I’d be more than happy for 98% of my drives, which are mostly long roads where I just need to steer and can’t even spice things up by going as fast as I can handle – been there, done that, got caught too often and paid enough for it (especially dignity with my employer who had to bail me out of that one as I need the car to get to work).
            So yeah, count me in for self driving cars. It’s boring already as it is, I’d rather be doing something else than staring at the road for hours on end to get my ass from A to B.

          2. @Joan
            Yes.
            I learned it, but since my first car with automatic I don’t like manual. The computer knows the RPM and shall do what necessary when I step the gas pedal to accelerate. It’s just more comfortable, especially in traffic jam situations.

          3. I hate automatic transmission. The computer doesn’t always know what to do. Unless there’s a switch to choose how it accelerates, but then you would just be flipping that switch into different positions instead of changing gears… And it’s slower, unless you have two clutches and double gears, but even then two paddles would be a lot better than automatic.

          4. I used to hate them too, but then after being forced to drive them all the time, realised you can get a lot of influence if not control, by deliberate use of the right foot. i.e. you kinda make “gestures” with the pedal and you can make it change up and down when you want… or use the shifter to put it in 3 2 or 1 or L if you want RPM or to engine brake.

        1. Driving can be entertaining when its deliberately done as such, with the correct vehicles, road and setup.
          99.9% of driving isn’t though.

          Difference between painting for art and painting a fence. Or something.

        1. You’re not supposed to test the limits of your car during normal commute with other innocent bystanders and bydrivers in the vicinity – unless you’re a sociopath.
          You do stuff like that on a proper track and I can’t see people go lap after lap in a self driving car on those, unless they get paid for it.

      1. That’s like saying that drawing is a menial task and digital cameras do it better. As far as “fear of the unknown” goes, most of the people I encounter who are excited about self driving cars are everyday people with non technical backgrounds. The people I know who are terrified of self driving cars are all engineers. Your comment is pretty much the opposite of reality.

  2. So if someone uses this and their car crashes into another, is the person who crashed because they were using this going to be pursued by the law the same as if they were driving drunk, because it sounds equally irresponsible to me.

    1. facts beg to differ, self driving cars are already, pr hour driven in a given area safer than any known human driver.
      that we have had to wait this long for a fatal accident with thousands of self driving cars on the roads already sort of speaks for itself, had you taken the same amount of human drivers for the same amount of time (several years) you can be sure there would be a fatality as well.

        1. probably not, any single brand or product may of course be flawed but we really wont know until someone other than himself gets a hold of it and tries it out in a controlled environment.

    1. I think it varies by state. Some vehicles do not require an “inside” rear-view mirror, such as trucks with large cargo boxes, but then they require multiple outside mirrors. Doesn’t one auto mfgr (Tesla?) use a rear view camera and view screen. They claim the outside mirror affects aerodynamics.

    2. Any car with a closed box in the back can’t per logic have a rear view mirror that will show the rear.
      The side mirrors are the important ones, and if you ask me I think the one on the drivers side is the only mandatory one.

  3. We don’t need self driving cars as badly as we need cars that refuse to take stupid instructions from bad drivers.

    Think about that, carefully, because it also opens up a door into the market that is a lot easier to make work and eventually lead to ubiquitous fully autonomous cars in a graduated manner that is manageable and minimises risk.

    The other point I’d like to make is I don’t care how smart your car is if it is a shitty little tinfoil box with wheels that will turn into a my coffin the moment anything bigger and dumber gets the better of it. The same applies to humans, I don’t ride with fools or in pathetic vehicles.

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

      1. Yes. During winter they brake on highways when some light snow drops from an overpass. I don’t want a car that doesn’t let me do something. What if that happened when i suddenly saw two cars coming side by side and had to act faster than I could think to get out of the way and back to the road after they passed. I might be dead in a car that thinks for me.

  4. I dont remotely get the point of these systems where you do nothing yet have to pay attention. :-/
    It just seems worse then normal. Is it like a personality test? Some people prefer to do absolutely nothing but pay attention while others would rather be doing, well, something?

    Not that I am against self-driving itself – and I do think its possible eventually. But merely this half-message stuff seems so weird.

      1. I remember reading about fully automated cockpits. To give the pilots something to do, they bring in a dog. The pilot’s new job is to feed the dog. The dog is trained to bite the pilot if he tries to touch the controls.

  5. Oh you know what the missing tech is we need for self driving cars? Eating, sleeping, washroom and cooking facilities, in sub RV size…. because when pedestrians can walk into traffic with impunity, they will…. it will take all day for a 5 mile trip.

  6. The problem no one has addressed is maintenance and planned obsolescence.

    While most of the people who will buy these vehicles and add-on units like GeoHotz’s system will have the means to properly take care of their cars. There are several things that are much better addressed when felt by the driver, such as when a vehicle has steering issue causing it to pull to one side. This can be caused by issues other than low tire pressure (low tires would be detected by the TPMS), like if the vehicle hits a curb or a large pothole, the alignment could cause it to pull, and no system I’ve read about addresses this at all.

    If a driver experiences this, they can make a note to get it fixed as soon as they can, but I’ve not seen a system that detects this, let alone notifies the driver, before it goes unaddressed to the point of unsafe tire wear.

    For add-on systems, How will they address a flat tire, by that I mean, after the driver has changed the tire, how do they tell the vehicle not to travel over the 50mph speed limit of the “skinny spare”?

    As these systems age, will the inspection requirements be more stringent for systems that require much less attentiveness to the road. You can debate whether people will stay atuned to the driving situation, but the reality is that those vehicles causing the majority of the problems will be “driven” by those who don’t pay attention as much and they may not do maintenance as well as necessary to insure other driver safety.

    The insurers will command a high premium to insure these systems for a long period until the tech can be improved and vetted, until the rates come back down, there will be many with the attitude of “I paid a lot for the priviledge to not have to pay attention, so I won’t.”

    Computers have been in cars for decades, and often last the life of the vehicle, but when they act up the performance suffers, sometimes to the point of not running. Now we have a mechanical machine with a computer in charge, and when that computer acts up we have death (reference to Tesla death here).

    Some may have already asked questions such as:
    What happens when the computer gets hacked and causes the car to drive up on the curb towards people like the guy did at Ohio State Univ. and else where?

    And many answer with “We’ll have a way to allow the computer to be overridden.”
    My counter question/scenario is this:
    It’s well known that mechanical parts such as the parking brake (AKA hand brake or emergency brake) will no longer work if left unsed for long periods. This braking mechanism would likely be part of the engineer’s plan for the driver to over-ride the system when needed.
    How can the driver over ride a system when the parts to over-ride it are non-functional?

    How can the computer test these systems (the parking brake is engaged through an entirely different system than the regular hydraulic braking system), or at least be notified of their non-functional status so it can know when to turn off “auto-drive mode”?

    An add-on system like GeoHotz’s system can’t test these things, a manufacturer’s system could easily integrate a few parts to do this, but I’ve not seen any of google’s systems that mention anything remotely close to testing the parking brake for functionality. Mainly since they have only been installed on vehicles that are not old enough to have non-functional parking brakes.

    The people who have non-functional safety parts, lets be honest, are likely the ones who will need them the most due to other mechanical issues their vehicle has.

    Ask yourself this: How often do non-techie people blow the dust out of their computers? and How often do those same people have computer problems that are more significant than the computer problems techie people have?

    People who care the less about something, need more help when the time comes.

    With the planned obsolescense tactics in place for the last century, combined with the certain failure known to happen with mechanical parts and the very highlikelihood that it won’t be cared for as well since people won’t have to pay attention to driving, we have a long way to go before these systems will give a warm fuzzy to the people who choose NOT to use them.

Leave a Reply to oodainCancel reply

Please be kind and respectful to help make the comments section excellent. (Comment Policy)

This site uses Akismet to reduce spam. Learn how your comment data is processed.