INV 11 – Robot Arm Wrestling – Leaf 1 – Proportional Control

Courtesy Aralani

Blogubarra – May ’11

OK Computer – I admit that i am writing this post today, inspired to write this down having seen all sorts of my colleagues come to grief after a few seconds of fun riding mechanical bulls.  These bulls seem to be all the rage in Texas at least.  I’ve just got back from there, and lets face it, felt a bit more than afraid of getting on one, putting myself in the hands of some kind of cowboy operator with a manual  controller, with the crowd screaming:-

“Make it go faster”

And the poor guy falling off after around 3 seconds.  Mechanical bull riding seems all about the embarrassment of novices rather than, actually having fun on the bull..  never mind. !

So this innovation is hypothetical, how would you create an unbeatable robot controller, which could not be beaten at arm wrestling, which was evenly matched in brain and brawn to it’s contestant, man or woman, with a beer in hand.

So innovation 11 is a cautious step down this pathway…

Mark I iRobot wrestler has a strong arm, a hand, and a controller managing the arm and it’s own winning strategy.  The initial controller is just a proportional controller, or servo regulator, and has a strategy relating to the setpoint or angle of the arm from the vertical.

I step up to the arm, have beer, see the lighted buttons, and plan to select “Right ARM” which is the hand I will use to wrestle the robot.

“Got any change mister ?” says the robot

“Sure”, I say

“Up for a challenge then… bet you can’t beat my arm !” says the robot.

“No worries mate, I wrestle tractors all the time, downunder.  You have no chance.” I say.

I sit down, adjust my Akubra, and then just put in the money, select right arm fast, grab the arm and see what happens.

The robot arm instantly starts humming and pushes up to the vertical position.  I let it, ‘cos I want to see what it does.  So the arm is just sitting there at vertical.  I push left, the restoring force on the arm increases directly proportional to the force I use to push it over.  The more I push, the more the robot arm leans over, but the more force I have to apply to get it to lean over.

At about 45 degrees to the left, the robot says ” ah I see you are stronger than I thought, I must apply more effort. ”

“No worries I say, go for it ”

The robot arm strategy engine figures that its proportional gain may not be enough since I got it over to 45 degrees, and I challenged it with “No Worries Mate”, something the robot didn’t understand.  But correctly guesses it had something to do with confidence, purely by the tone.

So suddenly the force increased and it had pushed my arm back to 22 degrees from centre.

I figured ok, proportional control being used here, because I can just keep on pushing and the robot can never win. I wondered what other tricks it had up it’s sleeve.

after 1 minute the robot says impatiently “Look buddy, I don’t have all day, so I am going to finish this off in about 30 seconds”

I quickly think, 30 seconds, the robot must be going to change the setpoint to 90 degrees against me.  So I have 30 seconds to figure out how to defeat this.

I figure, since the robot is currently set to drive the arm to the vertical before the ‘coup de grace’ all I have to do it trick it into driving the arm my direction, and then suddenly I can win.

“Hey Robot, I exclaim, in 10 seconds you will be defeated!”

Before the robot has time to reply, I gradually relax my arm and the robot arm pushes it to vertical, right up in the middle, and at that point there is no force from either me or the robot. I suddenly pull the robot arm past vertical to the point where it is about 10 degrees from winning, I move fast so that the momentum of the arm moving helps move it further before the restoring force catches up.  Since it is proportional control, and the robot hasn’t changed the setpoint, the robot controller immediately applies a restoring force to push the arm back to vertical, i.e. going my way, with full force.  I suddenly reverse direction and both the robot controller and I are both pushing the arm at full force in my direction.

“Geronimo” I exclaim, the robot arm moves rapidly past vertical in my direction and the momentum of the heavy arm and my force push it over further and further until it hits the table.

“Uncle !” says the robot.. – who taught this thing language ? who the hell is “Uncle” ?

“Mate !.. see thats how it is done. Cheers “.

Proportional Control

In proportional control, the controller generates an output or restoring force proportionally equal to the difference between the setpoint and the actual position multiplied by some gain factor.

Generally speaking proportional control is kind of ok, but depending on the disturbance pushing the controlled variable away from the setpoint, you will always have an error, since it is the error which creates the control output.  No error, equals no output.

Innovation 7 – Miles on Fire as Engineering

It was April 1988.  I was working at Costec in Sydney at the time, a company then designing and making custom simulators for the power industry focussed on the idea of it’s two founders. Each one different, using currently available technology, fit for purpose – to train operators, and difficult to create, hours of work, based on a simple idea.  Wonderful time that was.

My first inkling that there was that kind of special thing was when I went to a concert at the time, It was Miles Davis playing in Sydney.  One of two concerts I think.  I had heard some of his music before, but not much.  I had even bought some sheet music of his and tried to play it. Didn’t understand it.

At that concert, I did experience Miles Davis, and got to know him through his music and ways around the stage. The way he kind of orchestrated on the fly, getting through his ideas in such an innovative way, each time he played a number. Here was one of the most intense people I ever saw or met. Someone who seemed to have an idea, and then do his best to express it through the entire composition, components of which were his trumpet playing, others were his sidemen. Music has a quality that is free of language.  Music speaks straight into the mind of the listeners, his kind of music devoid of lyrics and words, it goes straight in, if you let it. I did.

So it has taken  about 25 years for me to figure out what it was I heard.  What I heard was the physical expression of a burning idea through music. I am sure that non of it that I heard would be repeatable, without the man at the centre of it.  Since it was not just about the music, it was also his attitude, stance, control, and direction of that idea, his total focus on it, in order to have his musicians play it the way he saw it.  There I saw a match and its consequent fire of an idea, a physical thing.  It is here now, and tomorrow it will be there as well, but it may be modified, discarded for a better one, developed. I could be better tomorrow, it could be irrelevant, but for sure, there is a strong will about that idea that has to get out.

Since that time, I have come to understand about innovation in a similar abstraction.

Take an idea and then think about how you would tell someone about that idea musically.  Would it be noise, a lone trumpet song, and orchestra playing, a symphonic movement, which one, first or last ? How solid is the idea.  If someone kicks it, will it last, will it just blow away in the wind ? How permanent is the idea or its expression.  How quick does it have to be made ? How long does it have to last ? What color is it ? How much of that color is in fashion ? When will it go out of fashion ?……

Would you care ?

Innovations are true to the idea, its clarity and purpose. They reflect the idea so that others can see it.  The innovation itself has a timeframe, a purpose and a reason. So engineering too, it full of innovation and surprise, resulting in physical things others can see and appreciate, for what they are worth, their repeatability, and value, and for how long they are around.