Three years ago I made a claw to
handle ‘radioactive fuel’ in a glove box. I returned to this last
winter, but soon found the glove box was clumsy and unsatisfying to use.
Then these photos reminded me that ‘remote manipulator arms’ rather
than glove boxes are used for working with strongly radioactive materials.
This was more ambitious, but the previous job I’d been doing had
involved too much time on the computer and I was feeling like a challenge
to keep me in the workshop for a while.
Remote manipulator arms are an ideal project
for a mad inventor
My initial prototype arms were moderately
encouraging, but it was Sam, my six year old grandson, who really
persuaded me they were worth pursuing. Even in a very rough form, Sam was
completely absorbed playing with it for 45 minutes. It only took me two
weeks to get this far, but making an arm strong enough and making the
whole machine practical took months, and it became a lot more complicated
than I ever imagined. In this respect its like a real nuclear reactor –
a simple idea that gets very complicated trying to make it foolproof.
I had initially assumed I would have two arms like
real nuclear manipulators, but I found this needed considerable skill. I
remembered Sam had held the prototype with both hands, so I tried a single
arm made to be held with both hands. This was much easier and more
satisfying to use.
Igradually
refined the linkages from the handle to the remote ‘arm’. The steel
tie rods got replaced with nylon cord to make everything more springy and
more resilient to being roughly handled. I was delighted to find it was
completely intuitive to multiply a movement (so a small rotation or tilt
of the arm’s handle makes the remote arm turn or tilt much further).
Discovering this was exciting. It gave me a new
appreciation of the human brain. Its amazing we are so competent doing
something as indirect as steering a car by rotating the steering wheel –
where the movement of our arms bears no simple relationship to the car’s
movement. Its because we have such amazingly sophisticated eye, brain,
muscle co-ordination. When we eat, we are completely unaware of what our
hands are doing – we only focus on what the tips of the knife and fork
are cutting or shovelling. The steering wheel or cutlery implements are
effectively bionic extensions of our arms. And then we are often hardly
conscious of what we are doing, our attention is directed on what we are
listening to or talking about while we eat or drive. (though driving is
obviously more demanding than eating because we can’t read or watch TV
while driving, which we do while eating.)
As I developed the arm, I became increasingly aware
of subtleties like its friction and springiness. They made a big
difference to how intuitive and satisfying the arm was to use. It made me
more aware of the limitations of computer human interfaces. I use an
expensive Wacom cintiq tablet for drawing sometimes, but even this has
time delays and parallax problems between the stylus and the ‘drawn’
line on the screen so I still prefer to draw on paper and scan the image.
Using computers to control physical machines involves
much more compromise. I’m sure remote surgery robots must have sorted
this out, but on a limited budget it doesn’t work well. Radio controlled
planes and helicopters are tricky to master mainly because the
relationship between the joystick and the movement of the craft involve
delays and are not proportional. They are certainly not instinctive like a
car’s steering wheel.
Even expensive robotic servo systems have small
delays between sensing and reacting. Its hard for a walking robot to react
fast enough just to keep its balance. I was amazed by the video of the Big
Dog walking robot recovering itself when slipping on ice. I thought
computer servo systems must have finally speeded up. But then I found its
recovery was mainly a result of cleverly placed big springs. Springs react
without any time delay. This encouraged me. Kids used to computer
interactivity might be impressed by the sensitivity and immediacy of the
direct mechanical linkage of my remote manipulator arm.
I fretted a lot about whether I could ever make the
arm strong enough to survive rough handling. Part of the answer was to
make it as easy as possible to use – people are much rougher if they are
frustrated! Kids can also be rough if they aren’t playing the machine,
just messing about. So I decided the arm had to be clamped when the
machine wasn’t in use. This had the added advantage of forcing people to
watch the instruction video – without clamping the arm, people get so
absorbed inserting the fuel pellets that they never glance at the screen.
Deciding the task to perform with the arm was
surprisingly easy. I had tried putting little cylinders into tubes in an
early prototype and the idea stuck. I was unsure about it until I worked
out a mechanism to get the cylinders back inside a box at the end of each
go. Psychologically, this seemed to make the fuel seem more dangerous and
the task more interesting.
