Have You Ever Crashed A Robot?

    They say you never forget your first one.

    2014, my first robot. FANUC AM 0iA, for welding some pipes together.

    I was barely trained. I didn't even know fully what a USER FRAME was, and the TCP was still a bit magical.

    I did the trajectory. Everything worked fine while I ran the program step by step. Fine and dandy. I was completely alone with the installation at the client, so I called in the project lead to do a demonstration. Because running the program step by step ONCE was enough, by my reckoning then.

    I start the robot. It welds ok for the start, and then smashed torch first into the fixture.
    Apparently I had a CNT100 J movement somewhere on the way. Was the first time I learned why you should pay attention to your CNTs and your FINEs. I bent the torch a little, so that was a lot of fun to redo. Cemented my knowledge of TCPs tho.

    Not quite "carnage," but it got very expensive. Many years ago, a Major Robot Customer decided to change their process for approving robot payloads, and had their own in-house engineer do an approval on each integrator's end effector design for approving build. Said engineer also provided the mass/Cog/Inertia values which were supposed to be entered into the robot controllers, no questions to be asked.

    This was when CoG and inertia analyses still had to be done from paper drawings, which leads to the disaster in the offing.

    The "trial run" of this new procedure was on a full-factory installation (~500 robots) of a new production line. Everything was installed, debugged, bought off, and in production... and a couple months later, things began breaking down everywhere. Bolts holding end effectors to robots were breaking, entire wrists fell off robots, all sorts of similar things.

    Eventually turned out that the person approving all the end effector designs was somehow working from drawing sets that had been re-scaled somewhere in the communications chain, and most of the robots were nearly carrying nearly 200% rated payload (mostly in CoG effects, not mass). At maximum speed/accel. And these robot controllers being... "low end"... they didn't throw any load or torque faults, they just kept running until the mechanicals fell apart.

    Ever seen a robot lose its entire wrist and keep trying to run like nothing happened, with a big spot-welding tool dangling by its cabling?

    In the end, the entire production line had to be re-tooled, at great expense. Many of the overloaded robots had to be replaced with heavier-payload models, which had shorter arms, which meant lots of tooling had to be re-located, and every cell was already packed to bursting due to tight floor space constraints....

    I have 4 IRB6 S2 robots and recently lost one at sea!!!

    We had excavated a pit in the floor and the robot was raised and lowered on a hydraulic table.

    We had heavy rain 1 weekend and flooded, and I came in to find she had drowned!!!

    I stripped her down and dried her out but she never ran right again, God bless her.

    I have replaced her with irb1400 IRC5 (what a learning curve that is!!!) and needless to say, the default platform position is raised!!!

    omg, not the water....

    that brings memories... some years ago there was case of a robot that was shipped across Canada. that is something that takes solid week even when things are going your way. in this case everything was wrong...

    winds managed to rip of tarps. water, snow, dust and road salt all made their way into every crevice. many surfaces were corroded, top of the cabinet had additional holes as integrator for some reason decided to attach some conduits to the top and side of the KRC using self-tapping screws. stains showed that there was trickle through every rivet, gland, CPS ports, disconnect knob etc. and water was pooling inside the cabinet. after a week or so of such harsh weather wire-mesh of the RF seal around the door was encrusted with salt and rust. nobody would tell that this was a new system. robot was somehow in slightly better shape.

    we learned later that upon arrival there was an attempt to clean it all dry it out but it was not enough. when the power was applied, most of the system was destroyed including some cables and connectors. but after all defective components were replaced, it worked. obviously this was not something that was covered by warranty.

    and do everything you can to keep the water out. always protect the robot and controller as if they are shipping across the ocean. and use insurance... if something goes wrong, you will be happy you paid for it. any repairs will be much much more costly.

    1) read pinned topic: READ FIRST...

    2) if you have an issue with robot, post question in the correct forum section... do NOT contact me directly

    3) read 1 and 2

    OK, I did crashed a tool and if the tool is part of the robot (check T06 and T07?) - ja, sorry: mea culpa!

    Crashing tools or robots (in my opion) is simply brains (after a 20h day with al lot of coffee ) much faster than physics?

    In other words:

    STOP at the right time

    saving minutes could cost you hours of time to recover

    lesson learned:

    destroying tool today could you cost 14 weeks to get replacement of the tool

    Yeah baby, welcome to the club :fine:

    Quote from MOM

    Crashing tools or robots (in my opion) is simply brains (after a 20h day with al lot of coffee ) much faster than physics?

    For sure.

    It's hard to stop when your deep into a problem and want a solution or at it for hours on end.

    The pixies are just waiting for that single opportunity for you to switch off and apply a simple fundamental error that turns your world upside down and leave you wanting to just :away:

    Heh. My most dramatic crash was when I made a I/O change that resulted in a stamping press biting the entire EOAT off the robot (and trimming the EOAT into several discrete pieces). Fortunately, in press room operations, protecting the die is one of the most critical concerns, so every EOAT that got between the dies was made of light aluminum tube designed to crush so easily that, even when I made that mistake, the dies were not in (much) danger of taking damage. And the robot was never programmed to put any of itself between the dies (not to protect the robot, but to avoid scuffing/chipping/marring the Class A die surface -- fixing the die from even minor damage could cost more than replacing the robot!)

    We had plenty of spare parts, so making a new EOAT and touching up to it only took an hour. But I darn near had heart failure when I saw the press close early. My life (and career) was flashing before my eyes....

    Most expensive crash wasn't mine, but a co-worker's. And illustrates the risks of high-speed teach modes -- he forgot he was in T2, and hit the Go button while standing right next to the robot. Fortunately, it moved away from him... and smashed into the brand-new ~$75000+ laser tracker right next to the robot on the other side. Before he could even release the deadman... SMASH.

    I remember once I had to add some new programs to a robot that grabbed injection molded plastic chairs and placed them on an array of 6 shelves for the chairs to cool. Shelves didn't have any sensors to tell the PLC which locations were occupied, this was all tracked in the (locked) PLC program.

    A day of running in an out of the cell manually removing the hot chairs, resetting the sequence on the HMI, and trying to coordinate with the Injection Mold operator, definitely resulted in the robot placing the fresh chairs in the location of an existing chair - sending them flying into the fence..

    Another funny memory of that job, grabbing a (still warm) chair off the racking, and sitting on it while teaching some of the place points - after about 40s my knees are above my ears and my bum is on the ground, the warm chair had turned into a Salvador Dali clock :grinning_face_with_smiling_eyes:

    I tried to move Home one of two robots in a shared 7th axis in a 2nd level, with no labels...
    Turned out that I moved the other robot that was dropping a part in a fixture :neutral_face:

    Gripper broke, all at 3 am -.- now days I make sure I'm moving the right one by checking multiple times and I write with a marker whenever something is no labeled or miss labeled.

Advertising from our partners