Wednesday, April 21, 2010

Twitch, Jr.

Guess what? I'm actually going to build something that isn't a motor or a motor controller. It's been a long time since I actually built a robot, but seeing a fair number of 2.007 robots coming to life makes me want to execute a good old-fashioned two-week robot build. Okay, so I'm not really going to build a whole robot, just a drivetrain. That's all I was ever good at, anyway. I've built many different types: 2WD, 4WD, 6WD, skid-steer, tank treads, mecanum wheels. But I've never built a linkage drive! What is a linkage drive? It's this:




This is the drive from FIRST Team 1565's 2008 robot, "Twitch...the missing link(age)." I am not affiliated in any way with Team 1565; I just think their linkage drive is awesome.

It doesn't have complete...uh oh time to make up a word...holonomicity? In other words, it can't go in any arbitrary direction at any time like mecanum wheels or a swerve drive. But it can rapidly swap between two perpendicular directions of travel, which is very practical. It has omni-wheels, but only just to allow it to turn and swap axes smoothly without skidding wheels sideways. Consequently, it can coast into turns as it swaps axes, which I think makes it incredibly fun to drive.

So, I'm going to make a (small) linkage-drive robot. Here's a quick video of the linkages in action:


 


Sorry for the awkward cropping...click the full-screen icon for a slightly better view. For a sense of scale, those are ~4" wheels. (They are really these awesome Vex omni-wheels, but I'm too lazy to add the rollers into this existing wheel model I found.)

Top View

The four identical wheel+gearbox+motor modules are mounted in plastic bearing blocks that rotate on upper and lower posts. The two long, diagonal links do most of the work, linking opposite corners to each other so that diagonal wheels are always parallel. The short links (one of which is redundant) couple the two sets of diagonal wheels in a strange and somewhat frightening way. (Frightening in that it comes kind-of close to singularity at the extremes.) That's why I left open the option of using two servos to drive the two diagonals independently. But in theory the entire mechanism has just one degree of freedom.

The gearmotors are 36mm, 20:1 BaneBots planetary sets. The controllers (shown as nondescript green blobs) are Pololu Trex modules. The structure is mostly an aluminum sandwich, with the plates and links being cut with an abrasive water jet. And yes, I sent the parts out despite the fact that MIT has seven waterjets on campus now. Cue rant: Of the seven, only one is open for general use and it's not currently working...plus sending it out is fairly cheap, believe it or not. (If you don't happen to go to a school or work at a company that likes to spend money on underutilized and difficult-to-maintain machine tools, here's one good place that will cut your cool robot parts quickly and cheaply.)

Future robot.

More parts arrive this week, so I should have a robot by next week. If you're wondering about the control...I have no idea how to do it yet. But I'm not really worried. I'll write it in Visual Basic just to annoy everyone. Shouldn't take that long. (Really.)

I just feel like there's something else I'm forgetting to do...oh right my thesis!

2 comments:

  1. What a life. 2.00thesis? This way more fun. Just brought back memories of playing around (watching other people play around?) in E60, and 3rd floor Koosh ball ambushes and things like that. By the way, yes I read this occasionally. Decided to say "hi" this time. Hi, Shane! -aD

    ReplyDelete
  2. hey, shane... great .... i m also use your link drive... i m very happy with it ... thanks

    ReplyDelete