Robotics & vision suggestions

Hey folks, lots of helpful information here but this is what has been dropped into my lap. We have aquired an older Fanuc robot and want to use it to pick up an item, and then place the item on another object when the larger item spins into position. We currently control our process with AB Contrologix so I plan on setting up Devicenet comms to the robot. Anyone have research suggestions for a vision system or other method that will guide the robot to exactly where the target has stopped? We are not using either Robots or vision systems currently so this will be a learning experience.

Any reason you decided to do Devicenet instead of Ethernet/IP? I know we do Fanuc from ControlLogix vie Ethernet/IP regularly (though I don't know the details).

As for vision systems, you really just need to pick between Overpriced Cognex, or any other brand. Competitors include Sick, Dalsa, PPT, Keyence and more. I would find out the requirements of your application and then get bids from the 4 non-cognex companies and pick the cheapest one. Things to consider are how far away the object will be from the camera, if that distance is fixed or you need a large focal range, will the target be moving (and how fast), how much time you can have between needing to trigger and having a result.

One big thing is if you can run the camera using a fast communication protocol or if you have to add a discrete IO board. You can save big money by getting the profinet or powerlink camera and dropping the IO board (I don't think these are options for ControlLogix though).

I'm with CapinWinky, Cognex has a lot of market share (and attitude) but there are other competitors just as good for less. I've used Datalogic (PPT) and Keyence with good results. Datalogic had a more extensive toolset and they talk Ethernet/IP. You may also want to involve an integrator depending on your timeframe.

Some experience but not a robotics expert

Maxnukem said:

Anyone have research suggestions for a vision system or other method that will guide the robot to exactly where the target has stopped? We are not using either Robots or vision systems currently so this will be a learning experience.

Click to expand...

A vision system may not necessarily be required if the incoming object's dimensions are consistent and it can be repeatedly positioned in the correct orientation for a 'pick' by the robot and the same conditions apply to the object on the 'place' side of the operation. If this is true you could use discrete sensors (PEs, PROX switches, etc.) and actuators (solenoids, air stops, whatever).

If, the picked item is randomly located then perhaps some actuators could be used to force it into some known orientation. Think pushing the object against a rail or into a corner. Judicious use of sensors could then inform the robot as to which of several pick routines to use.