Emerald Group Publishing Limited
Copyright © 2000, MCB UP Limited
An idea whose time has come
An idea whose time has come
The theme of this issue is "robot accuracy" and we cover the numerous reasons why inaccuracies occur, ways of minimising errors and the measures being taken by at least one manufacturer to address this most important, but previously rather ignored, aspect of robotics (see "High accuracy positioning system..." on page 274).
Robots have always been pretty hot on repeatability. About 15 years ago I was involved in some benchmarking trials of various industrial robots and one test was to position the robot's end-effector against a micrometer dial gauge, teach that point as a position and then waggle the robot arm around all over the place under either teach pendant or program control and finish with the command to move back to the taught position at full speed. I am not sure who was sweating the most between the robot salesman who knew that a bulk order depended on the results of the test, or the service engineer with visions of his expensive dial gauge going flying across the factory floor. Either way both were soon able to relax because the robot returned perfectly and the dial gauge survived to measure another day. Unfortunately the same cannot be said of accuracy or anything requiring the movement to a 3D co-ordinate specified (as opposed to pre-taught) point or along a similarly specified path. A robot with pen in hand that was instructed to draw a castellated line with sharp right-angled corners would instead produce a crude approximation with rounded corners and wavy edges. Even worse, a path composed of relative movements in 3D space accumulated compounded errors to produce a banana shaped trajectory that could literally be out by several hundred millimetres after a run of a meter or less.
Everyone makes mistakes but it is a good idea not to make a habit of it, and even better if we can learn from our mistakes so that they are not repeated. Unfortunately robots tend to make the same mistakes over and over again, which has always seemed to me to be a bit careless. I am referring to the mistakes or errors that occur when a robot deviates from its intended path. The reasons for the error will be entwined in the tangled web of kinematics and control theory but the point is that the robot will know where it is meant to go and will know the errors that it made. However the next time it repeats the program it will make the same mistakes as if running the path for the first time.
Something clearly needs to be done but it is to the industry's shame that this whole issue of 3D accuracy has been swept under the carpet for far too long. In their defence robot manufacturers have had other priorities and we have all benefited from improved reliability, precision, speed and load carrying capacity. However in recent years the bulge under the carpet has been growing and more than a few have tripped up over it, and this has lead (finally) to this issue being taken more seriously.
Two of our contributions make a common statement and I will make it a third by suggesting that all purchasers of robots insist that robot suppliers provide them with detailed specifications of accuracy based on ISO and ANSI standards. In the end it will only be those robot manufacturers who ignore these requests that will lose out.