Closing a robot gripper and stalling a motor when grasping an object.

Martin_H

An item I plan to plug into my hacked Scribbler's headers is a small servo controlled robot gripper. That way Scribbler will be able to grab and carry small items. It occurs to me that if I tell the gripper to fully close, it will close and grab an object, but the motor will also stall.

Is this a reasonable way to grab an object, or is it bad it to stall the motor on a servo? I imagine current consumption would go up and I am not sure how much current Scribbler can supply.

I suppose a touch sensor in the gripper could tell when it has grasped an object. Incrementally closing the gripper until the sensor triggers would allow me to determine the servo hold position. But that seems fairly complicated, would consume another header, and even being off by a little would likely stall the motor again.

Does anyone know of any strategies with gripper control?

An alternate strategy is to add local brains to the gripper where 1 on the I/O pin means close, and 0 means open. The local brains could monitor the position and touch sensor in an effort to prevent a stall. But that's even more complicated.

erco

In addition to a position sensor, many robotic grippers have force feedback. This is essentially a torsion spring somewhere in the geartrain coupled to a pot. As the gripper closes on an object and continues to drive itself closed, the gripper jaws squeeze with increasing spring-loaded force. You could rework a standard servo to operate something like that with force feedback only, so you would spring load the pot in one direction and would only move as the grip force increased. In that case, you would want to add a limit switch when your gripper opened fully. So to grasp something, you drive the gripper servo open to the limit switch, then close it to a certain pulsout measurement corresponding to a certain gripping force and let the existing servo electronics handle the sensing.

Alternatively, add some conductive foam inside one of your gripper jaws and monitor the resistance across that foam (RCtime). When the resistance changes, you have made contact with something.

Probably a good application for a servo-saver coupling to avoid stripping gears.

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
·"If you build it, they will come."

Post Edited (erco) : 3/24/2010 1:36:57 AM GMT

Martin_H

Thanks Erco, force feedback sounds ideal, but I'd rather buy a servo set up to do it. Hacking one sounds a bit too tricky.

The complexity of an electronic approach shows the appeal of a mechanical approach to solve this problem. If I understand the servo savers, the servo isn't driving the linkage directly, but a spring. When the linkage stalls the spring is compressed to absorb the extra travel distance. Looking at the Parallax Boebot gripper it looks like it translates extra closing force into lifting force which is even more clever, but there's no way I could attach that to Scribbler.

So the servo savers look interesting. Hopefully I can find one that would work for the servo/gripper combination I have.

Post Edited (Martin_H) : 3/24/2010 11:32:30 PM GMT