[ Home ] [ Up ] [ 1st Generation ] [ 2nd Generation ] [ 3rd Generation ]

The Genesis of the Universal Planar Manipulator -- from Ancient Times to Now

Version 1 (1996)

In the beginning there was a baking tray shaken around by voice coils yanked out of old disk drives (thanks to Eric Paulos for pokin' around and finding them at HMR). The baking tray was cheap and wobbly, the base was made of corrugated wood (John, where did you find that?). The motor shafts (thin brass sticks) doubled as the tray's suspension and as lateral flexures. Despite the humble beginnings, we still managed to swirl and translate pennies around.

Version 2 (1997-8)

Then John (not an apostle) said, let the baking tray be used for cakes. We'll replace it by a 8"x8"x1/8" sheet of formica, a robust plastic used on kitchentops.

The whole system: table, suspension, motors, camera, signal generation (with 2 soundblaster cards), controlling motors -- was put together. We managed to do closed-loop 1-part motion quite well. But when it came to moving more than one part...

Version 3 (1998-9)

We realized more motor power was needed. Formica at that thickness was not going to stay rigid, and the disk drive motors started to fall apart. So thus spake John (which by now was becoming a Saint): I command you to replace formica by Corian, a more robust kitchentop plastic (yeah, we like baking, cooking, etc.). So the plate became a 12"x12"x1/4" sheet of that heavenly material. The disk drive motors (may their souls rest in peace) were replaced by 50 lbf voice coils. Much work went into designing and manufacturing their holding braces and the motor shafts' flexural suspension. The cheap corrugated wood based was replaced by a solid aluminum breadboard. Soundblasters with all their IRQ conflicts were tossed out: use microcontrollers, spake he. Evening came and then morning—that was the third prototype:

Version 4 (1999-2000)

A whole lotta shakin' was going on, but the Corian, a solid material, was still wobbling up and down like the ill-fated Tacoma bridge. And so was the solid aluminum base. One day there fell from the heavens an idea (Joe Gavazza from the machine shop was the messenger): manah! manah! I mean, honeycomb! honeycomb! Is it not this composite material which maximizes stiffness by weight? And thus came forth yet another version of our prototype: the plate became a 1" thick honeycomb tile (thanks to Teklam, inc.) and the base became a sturdy Newport optical table (thanks to Heather Yaros and the Radke Lab), which internally is also a honeycomb.

Then we looked at what we had done, and it was good. But we're planning to go further: