ORACLE - Our First RobotORACLE - Our First Robot
The Vision
(Picture above is the model I built at the time).
In my mind, I could see a small cutter skimming over the sheep like a low flying aircraft, hugging the hills and valleys, slicing the wool away in white waves of fleece. It was guided by a computer which precisely remembered where it had cut, and simply turned it round at the end of each blow, and made it skim along a new path exactly alongside the previous one. It was gentle, yet precisely efficient. I naively thought it could run at up to 50 or even 75 centimetres per second, and shear the whole sheep in about a minute of continuous movement. The sheep would have to be rolled over now and again of course, and that would take a little time. Still, it would be much faster than manual shearing and effortless.
The sheep would have to keep still. But then, all the sheep I had seen on the cradles at Muresk and Geelong had kept remarkably still. If the sheep struggled, the robot could backtrack a little and resume once the sheep was quiet.
The resistance sensing technique seemed to be full of potential, and as a sense of touch it would allow the computer to 'discover' the exact shape of the sheep progressively. The computer would start shearing with a reference path for the first blow. Then, as each blow was shorn, the computer would gather a more and more comprehensive 'map' of the sheep.
Genesis of ORACLE
Once the review of the PATS and CSIRO machines was completed, Roy Leslie returned to a study of sheep restraint devices while I developed the control system ideas. Jim Blair and Prof A-W compiled an application for $75,000 to start building a machine to test these ideas. Months of delicate negotiations followed. CSIRO were reluctant to include us, but the Corporation were strongly in favour of our contribution. By October 1977, CSIRO had emerged as project leaders, under David Henshaw's direction.
(Hydraulic servo demonstration - Roy Leslie 1977. This became part of ORACLE)
Roy Leslie and myself (part-time) were to begin work on a computer controlled machine for experimental shearing. Yet CSIRO regarded us as a sub-contractor for the design of a computer control system for their own machine. By now, they appreciated the ease of changing software in a computer control system. On the other hand, we thought that the mechanical design concept which CSIRO had adopted was a major constraint; unless this was removed, the full potential of computer control could never be realized. We also knew that we needed our own experimental equipment - no one knew enough to design a computer control system without having to experiment with it first. We had to persuade CSIRO to let us build a test rig for our own use.
Even if David had acquiesced, it would have been clear to his CSIRO colleagues that our machine would soon eclipse anything which could be usefully achieved at Geelong if we were allowed to proceed. Even though the Corporation had nominated CSIRO as project leaders, it might be a title in name only.
David Henshaw visited us for the first time in December and we were able at last to give him a close look at our ideas. We showed him a trial hydraulic actuator Roy Leslie had built as soon as we had received word to proceed. It was awesomely powerful and fast as it hammered from one end of its stroke to the other in a fraction of a second.
We took him to the darkened computer room to watch a simulated shearing robot shearing simulated sheep, tracing delicate patterns across the glowing green screens. We looked at a real sheep, particularly some of the harder areas for shearing like the belly and crutch. We laid it upside down, pulling legs this way and that; then on its side, curling its neck around, for Roy had already realized that manipulating the sheep's body was the only avenue towards complete shearing. By the end of the day David seemed to be catching some of our enthusiasm. It was as if he realized then which way the project had to go.
David revealed his new design for a low power cutter. This was good news for us. A conventional cutter needs up to 500 W of power and the only available motors were heavy and bulky. We could lighten the cutter mechanism for our robot and the smaller size would make it easier to design. So we commissioned David to build one for us to use on our rig.
By the end of February 1978, CSIRO had conceded that we needed to build
a robot. Our concession was to focus initial work on the back and side
of the sheep - the same area as CSIRO were shearing. This turned out to
be a near fatal mistake, one which was to cause us many technical problems
later. In retrospect, we ought to have designed a robot capable of shearing
other areas too, and then simply used it initially 'as directed'.
|Tour Page|James Trevelyan's
Page| UWA Home page|
Web site design: James Trevelyan jamest@mech.uwa.edu.au
September 1997