Research is a lot of fun, but lab work can sometimes be really repetitive and demands a great deal of focus. Some of the tasks I do on a daily basis would benefit greatly from automation, but lab equipment in general, especially automation is extremely costly. Or at least that's the way it used to be, there are now an increasing number of open source/DIY equipment out there (http://openpcr.org/, http://www.chaibio.com/, http://www.opentrons.com/). I'm really excited about the opentrons liquid handling robot (hope they can deliver on time) and thought immediately that coupling this machine to other machines in the lab would open up tons of new possibilities. Unfortunately robot arms are also pretty expensive (http://www.strobotics.com/prices.htm). The plan now is simple, I would like to build a robotic arm that can transfer micoplates between the liquid handling robot and a plate reader (we have a spectramax M5 which has a rather simple api for automation).
Finding the right arm
I wanted to build as much as possible on some other well made project, so I don't have to work a lot, I would have something functional that can be improved upon fast and there would be people who I can ask for help. Robotic arms with 4-5 or even 6 axes of freedom are common, but they are all controlled by the usual small servos (the µarm is probably the best: http://www.ufactory.cc/), which I personally don't like too much. This guy is the only exception I found, but there is nothing else shared than the video:
Pity, its super cool!
So I started thinking about using a delta robot. I had my trusty Kossel type 3d printer for a year now, and it is still pure magic! I knew a delta would get the job done, and started to look for ones with the more common design (arms attached to horns rather than linear rails on columns). There are again a lot machines running servos, but I finally stumbled upon the FirePick delta:
Extremely neatly built and well documented. It runs Marlin so it's already a bit familiar.
Things to change for my application:
Arm lengths: I will need a machine with a larger work area. This way the resolution is lower, but it should still be ok.
Magnetic ball joints: This is not really a change, they are also moving away from the traxxas joints. I had the aerostock design (http://www.thingiverse.com/thing:93042 and http://www.thingiverse.com/thing:95258) on my printer and I really love it. I think I can't use their end effector, because the gripper will be too large to hot swap.