Tuesday, September 21, 2010

What do you exactly do?

I've had to explain this too many times to too many people. And the answer is not a simple "Robotics" anymore. So I decided to write a blog post about it. Next time someone asks me this question, they get the link. :-) (Phd & scientists are all about being lazy.) Except since I like to blabber, this explanation is gonna be long winded and mostly non-scientific.

Imagine a machine which you can accurately control. It has been shaped like a human arm and has a pointy tip at the end. You can control exactly where that tip reaches, right down to under a millimeter. Since the machine tip can be accurately placed, it means that you can figure out when it is placed incorrectly. Or is placed a little off. So you can use this machine not just to place something accurately, but also to check if what has been placed is in the right spot or not. This is how measuring machines work.

Next, observe when you have to pick up a heavy container filled with water. Using just one hand, you can lift it up. But it is probably too much strain and a little hard to balance. Rather than a container, think of it as lifting a chair. Now instead, if you use both hands, not only can you lift the chair (or container), you are also guaranteed that chair is held correctly (in case of the container, the water will not spill). If two people are lifting the chair, controlling the angle of the chair is even easier. And you don't need heavy muscles for that.

A suspension bridge is similar in concept (at a very basic level).

What I just described is analogous to something called as a "parallel mechanism". I won't go into more details.

Now recall a string puppeteer. He controls his puppets with a string, and can give an amazing range of motions to the puppet. What he does is move the strings - only up and down - to give motion to the puppet. The length of the string determines how the puppets' limbs move. This is also a parallel mechanism, except here the strings are doing the work of the arms in doing the "lifting".

Now, instead of the puppet, some strings were attached to your hands. A string or two above your elbow and another string or two on your forearms. If someone pulled the strings to your forearms, and kept those to your elbow taut, only your elbow joint would move. On the other hand, if you moved your elbow joint, I could just measure the length of the string (how much it is getting pulled). Since I know that string length change corresponds to hand movement, I could just keep track of the string lengths and figure out what kinds of motions are those that can be observed in healthy persons.
Essentially, this above paragraph describes what we want to do. But just to be sure, we want to add accelerometers, force sensors, IR camera based motion capture systems and other possible stuff to verify and improve on the estimates obtained by measuring those "strings" or wires.

Does that make sense?