Homemade Industrial Robot – First Accuracy and Repeatability Tests

This was the first computer controlled bench test of the first four joints of the arm, to see how repeatable and accurate its drive systems were.
The arm was only being moved through part of its range to prevent pinching of the wiring harness, as the harness didn’t yet have proper frame supports added. The drive components also hadn’t been tensioned or adjusted, so some looseness was expected. The drive cables for the wrist and gripper were also not in use yet so were just strapped to the forearm motor to keep them out of the way.
A metal scribe taped to the end of the forearm joint shaft acted as a basic ‘pointer’, and scrap timber with a needle point screw pointing upwards acted as fixed reference point.
The servos were controlled by Mach3 software using a simple G-Code script containing coordinates initially derived from manual positioning of the pointer. The plan is to develop proper motion strategy algorithms later on so the arm can be programmed in proper XYZYPR mode.
All in all I’m quite happy with the early results, and repeatability appears to be in the 1-2mm category with no load attached. After proper tuning I hope to reduce this to less than 1mm across the entire range of motion.
An issue with noticeable hysteresis in the elbow joint was seen, possibly because of a lack of lubrication and tension, but more likely because the motor mounting bolts were not yet torqued so the mounting fixture may have had some play in it. This will be looked at later. Fortunately the hysteresis appeared consistent, and repeatability was excellent when the same approach vector was used each time.
PS the ‘soundtrack’ of birds in the background is courtesy of the bush wildlife in Sydney’s northern suburbs.

From:
Date: November 29, 2017