Gantry robot solution - safety

    I'm looking for some ideas. We have a Yushin gantry robot that has a rather large EOAT (about 14"x20") that picks an array of inserts (12 at a time). The inserts are staged on trays by an operator, the trays pushed into place and locked. There is a substantial opening (about 20" wide x 4-6" tall) which the operators can reach their arm in the cell, and the robot picks the insert right at that opening. The integrator obviously didn't think out the safety hazards on this one (it was installed about 2015, before I worked here). The Yushin is single channel safety.


    I have added guard doors to block this opening while the robot tends, but am unsure as to the best, yet somewhat cost effective method into ensuring the guards are in the down position when the robot is tending. I was thinking of adding locking interlocks, but don't know the best method to tie into the existing safety circuitry (there is only 1 input). Any good suggestions? I was somewhat thinking numerous limit switches, but that will become very cumbersome.


    Also note: the Yushin doesn't have a "teach mode' per say, so if the fence is open, you cannot move the robot in any way, plus it kills all 24 volt power (even to signal switches). The tray areas need to be accessible to pull trays out and push them in.

    This robot sounds like it does not adhere to RIA safety standards. Whoever decided to buy it should be forced to put their own skins on the line and operate it in perpetuity.


    As far as trying to make the operator opening safe, the trick is finding a reliable way to measure the robot's actual position, relative to the opening. With a gantry robot, this gets hard b/c the motion envelope is so much larger.


    The classical way to do this would be to add a physical switch for each axis (or at least the primary axes) and "trip dogs" that only depress a switch when that axis is in a clear-of-operator state (or go the other way, and actuate on a not-clear-of-operator state). One key is to use normally-closed logic -- that is, if someone unplugs a switch or cuts a cable, that reads as an unsafe condition and forces the robot to stop. Wire these switches in series with each other, and in parrallel to the guard doors. Wire that assembly in series with the existing safety circuit. This should create a situation where the robot keeps running as long as one safety is satisfied (ie, either the doors are closed, or the robot is too far away to strike the operator), but any time both circuits are opened, the robot gets killed.


    Another way might be to add some light screens or maybe a SICK or Keyence area scanner that can detect when the robot is entering the volume where the operator's arms might be. Again, wire it in parallel with the door guards, and in series with the safety circuit.

    You are correct, it definitely does not adhere to RIA standards, for a half dozen reasons too. :waffen100: I am going forward with the limit switch option as well as a dedicated safety controller. I will be putting limit switches on all 3 linear axis and will monitor them that way. I think its the only viable solution (outside of having the whole cell redone).


    I originally was thinking of a safety scanner, but because the robot is so close to the operator working area, I don't think I will be able to reduce risk adequately. In my experience, the safety scanners experience some noise when reflecting off aluminum, so I don't think I could provide the adequate protection the operators should have.


    I cringe when I look at our older equipment and the lack of forethought when it comes to safety. Interestingly, even though we have a couple of gantry robots that are less than 5 years old, they all contain single channel safety circuits as well, which surprised me. Worse yet, the same integrator actually just wired the safety interlocks in a (literal) series on one of the cells.

    Another option instead of using multiple limit switches on different axes is to have a single "home" safety limit switch that the EOAT "parks" against while the operator is loading/unloading the cell. Wire as previous answer that park OR gate switch need to be made to hold off the E-stop circuit. You could wire the switches into one of the many monitored safety relays available so at least your additions would be dual channel.

    Unfortunately the dynamics of the job don't allow the robot to sit idle for any given time (operators load a total of about 192 inserts on a tray at a time and then leave that aspect of the robot unattended) and press cycle time consistency is super important. :frowning_face: In this case limit switches to monitor the axis position will be the easiest I believe.


    I plan on having limit switches for the robot at the upper position and forward-most position (well beyond physical reach through the gap). These will be safe positions. If the robot is not in either of them, I will also have limit switches at each tray, so the limit must be on and the gate closed, or a fence fault will occur.



    Quote from RoboWeld


    Another option instead of using multiple limit switches on different axes is to have a single "home" safety limit switch that the EOAT "parks" against while the operator is loading/unloading the cell. Wire as previous answer that park OR gate switch need to be made to hold off the E-stop circuit. You could wire the switches into one of the many monitored safety relays available so at least your additions would be dual channel.

Advertising from our partners