Wednesday, October 17, 2018

Disney is Working on Robotic Stuntmen

An interesting little story on Machine Design about the effort to make autonomous robots that can take the place of stunt actors in dangerous scenes.

The story follows the development of a simple "BRICK" (Binary Robotic Inertially Controlled bricK); through a Parkour robot that was tested on an air hockey table; to a 7' tall stick that flips end over end; to a stick that bends, somersaults, and always lands on its "back" (sort of a simplified version of human bending their knees up and somersaulting), and finally to the anthropomorphic robot seen in the 40 second video above (which resembles the title character in Avengers: Age of Ultron).

It's a straightforward progression; methodically adding complexity until finally the humanoid looking robot is flung from a wire cable and maneuvers with "no strings on me".  I can see simple tasks like these being done by robots, but I'm still skeptical of the "robots will take all the jobs" idea that's so prevalent. 


  1. Putting robots into those situations is always a good idea. Thus making robots that will be destroyed by kinetic damage, melted, etc. would seem to be a good way forward if there is money to do something like that. Melting stuntmen creates problems with the screen actor's guild or whoever represents them.

  2. It's when, not if. Eventually robots and AI will reach the level where they can do just about anything humans can do.....often better than we can. If we're lucky they'll be benevolent and benign like Cmdr Data from SNG. More likely they'll be dangerous, more like Skynet/Terminator/The Forbin Project.

  3. Aesop:
    1) That's not a "robot", it's simply a machine that can do 1 or 3 human-y things whilst being flung through the air.

    2) Some reality from 20+ years behind the camera:
    A 20-40 person team of construction grips built the contraptions for that set up, over several days. Possibly even several different teams.
    Another 5-10 are standing by for the actual shot.
    4-5 electricians lighted it.
    Another 6-10 SFX personnel operated the mechanical stunt dummy (not "robot").
    It took 10 more to build it. And probably for months, and probably the 4-10 other ones, for when that one gets broken, fails, or just shits the bed. (Read the behind-the-scenes on Jaws and the "revolutionary" mechanical shark sometime.)
    You had a crane operator pulling down $100/hr. Maybe 2-3.
    Throw in security guards, 3-6, 24/7, just to watch the site.
    2-5 person camera crew per camera filming that.
    (And 1-4 cameras, including maybe even a remote GoPro type inside the maquette's head.)
    Assistant director.
    A couple of production assistants.
    A medic for that off-site crew.
    Wardrobe person and assistant, as needed, if you put so much as a hat or shoes on it.
    5-20 additional personnel for miscellany I'm not even thinking of.
    10-500 CG, editing, etc. persons for post-production.
    Accounting, payroll, Teamster drivers, craft service, etc. etc. etc.

    In short, your animatronic dummy, for a 5 second shot, employed over 100 people for weeks to months, and possibly five times that many, some for half a year.

    Yeah, "robots" are taking over Hollywood jobs.
    (So, ask yourself, did the robot shark in Jaws create less jobs at Universal from 1976-present, or more...? I'll wait.)

    I spent most of a week, with a 50-person crew doing CGI for Matrix II & III, just to get reference film for CGI use, with 40+ of us at 40,000', doing zero-G Vomit Comet dives, two hops a day, just for background stunts.

    Absolute most fun I ever had at work in Hollywood, and they paid us to do astronaut training NASA used to spend millions of dollars to conduct. (Our entire air crew were former NASA Vomit Comet pilots and flight crew.) And that gig didn't exist in Hollywood from 1900-2000.

    Robotic stunt dummies?

    You just tripled production costs, and employed another 100 people, on any movie they're used on, forever.

    Please, invent MORE; IATSE and SAG will cream their pants and buy you steak & lobster dinners, forever. And the stunt guys will soak up the easy gigs, and they'll save the spendy machines for the stupid-dangerous dumbshit for productions with more money than brains. So no more jumping off 500' dams, or dangling from cables to get into blown-open 747 cockpits. Boo frickin' hoo.

    Back in the day, the pocket-protector dorks promised that "computers would reduce paper records" too.
    I doodled and drew for 18 years of growing up on just the reams of discarded paper that they "reduced" for my mom's accounting position with a Fortune 50 company. (This was back before there was an entire industry to store, shred, and recycle it, that didn't exist in 1970 in any way whatsoever.)

    And the reduction in a handful of accounting staff was offset by the addition of hundreds to tens of thousands of what we now refer to as "IT staff", from programmers to sysops, times infinity.


    Telling me tech is going to reduce jobs is like telling me lawyers will reduce lawsuits.

    Stop, you're killing me, and my sides hurt.

  4. Robots will indeed take over some jobs.

    Fast food restaurants will find that a robot will make their products faster and more uniformly at lower cost than humans. Although those robots will not be likely to apply the "special sauce" like those "urban youfs" who don't want to be working in the first place.

    "Research" jobs which consist of looking up previously published material will be taken over by AI, which will do a far better and faster and more thorough job of finding every available option based on key words. A small number of proof readers will be needed to make sure the results look right, but "legal aides" will almost disappear. And the same goes for a LARGE number of similar white collar jobs.

    Hazardous jobs in construction will be taken over by robots as soon as they're cost effective. Construction work tends to be hard and dirty, and as a result the snowflakes don't want to do it. That means they get the "Hey, y'all, watch this!" crowd, with the resulting injuries and deaths, followed by "help" from the government. They also get significant shrinkage from that crowd, so there's an additional reason to go robotic. Some of those jobs will lag - a crane robot on a highway construction site will be one of the last, due to the difficulty of properly programming same to address ALL the hazards of that job. And there will be a need for an onsite supervisor to keep an eye out for malfunctioning robots and unexpected "errors" for quite some time into the future.

    Anyone entering the workforce in the next five to ten years would do well to take at least one robotics and one AI course, to understand where things are headed.

  5. Aesop:
    Not Anon, Google just refuses to plug me in here, so I sign my posts at the top.

    And I expanded that thought somewhat, by way of underlining SiG's wisdom in being skeptical of tech taking jobs.

  6. Well, they tried cadavers, but, this lost the life-like appearance.