I'm in the early stages of planing a robot that will transfer one item ~300gm in and out of 5 different tanks of fluids. The tanks will be cylindrical ~100mm dia, ~200mm deep, and packed within a few cm of each other in a box. I'll require a minimum speed in all directions of 10cm/s, 20cm/s preferred. I estimate that accuracy during the run of ~10mm in all directions will be adequate, so that should be easy
I'd like the box/packaging to be small enough to store in the closet when not in use. If it is broken down for storage then it needs auto-calibration when reassembled.
The robot will not need to release and reacquire the payload during a run. The payload can be manually attached/detached at the start/end of a run if necessary although automated attach/detach would be a 'nice to have'.
I've never tried anything like this before and I think my options are a 2D gantry + winch, a 4-wire delta (I'm not sure that's the correct name, but hopefully you know what I mean) or a 3D arm
My thinking is:
gantry pros - relatively low torque (read cheap) stepper motors for translation, maybe the winch will need higher torque but not much more.
gantry cons - a pretty frame may be expensive, requires belts to get speed needed
delta pros - less physical structure required? identical motors throughout
delta cons - higher torque to hold load if its at the outside edge of the working space? Reducing torque load requires a structure taller than the working volume, which is annoying to store
arm pros - smallest physical package?
arm cons - requires servo motors with high torque to hold payload at maximum range. I estimate maximum range to be ~40cm allowing for required vertical motion.
Have I missed anything? Am I overestimating any of the cons?
Based on that, which option is likely to be cheapest, lowest maintenance requirements and look nicest? Where should I focus my research?
Arm will also introduce some ‘swing’ (and mech stress) which could possibly slow down your arrival at each target as the movement is settled.
Especially at outer reaches of the arm.
The delta and gantry are inherently self-stable structures if designed correctly (like a closed loop system) with the right materials, but the open ended elements of the arm are ‘the giveaway’, and require feedback & compensation to achieve similar results.
The table influence only comes into play if there is likely to be some distortion, deflection or slippage between the mechanism, target or table.
Delta needs a lot of vertical space to get a reasonable working volume. I think a gantry might be best for you. Particularly if these tanks are all in one line - then you only need a 2-axis robot.
MorganS:
Delta needs a lot of vertical space to get a reasonable working volume. I think a gantry might be best for you. Particularly if these tanks are all in one line - then you only need a 2-axis robot.
A 4-wire "delta" like a spidercam can use basically the whole footprint of the available area can't it? That would just make height an issue, which may not even be that bad compared to the structure of a gantry (I think - I'm doing some rough drawings to get a feel for dimensions).
Linear is great idea! I'm embarrassed I didn't think of it. Although I think once I add all the tanks plus space for a rinse station and a landing pad for the payload, it may get a bit too long, although I'll definitely investigate further. Thanks for the idea.
If the tanks are arranged in a linear row and you have a gantry that spans all 3 then you would only need two motors - one to move the traveler back and forth and one to lift and lower the payload. Think of the machines that are used for unloading containers from ships (but smaller ).
I think a gantry system would require a lot less headroom than a Delta system and would be a lot easier to make than an arm with equal stability. the gantry rails could be made from two pieces of plastic or aluminium angle that is readily available in DIY stores.
If you use an IR beam-break system to identify locations you should be able to do the whole thing with cheap geared DC motors.
The linear gantry system sounds like the most robust system to me. The winch is probably the weak spot. To throw a curved ball into this have you considered a system where the linear movement of the gantry is mechanically linked to the vertical movement of the item?
This would of course be far harder to design but would require no calibration and would be a thing to be admired.
The vertical movement need not be constant with the horizontal motion so you could get a rapid rise and fall to dunk the item into the liquid followed by little vertical movement whilst travelling between tanks.
If you think about it this kind of thing was being done in manufacturing in the steam age.
Here is a brilliant resource showing mechanical movements. The icons in colour are animated;
A few years ago, i built an x-y gantry (chunky belt steppers) that carried two linear actuators (operating vertically), along with the delivery nozzles for Cavro ‘syringe’ pumps.
It wasn’t all that hard to build, and was running on a Mega - with all sorts of speed ramping, limits, presets and other bells and whistles.
The work area was approx 1500x1000 mm, due to the specific project, the Z-axis was only around 150 mm.
Positioning accuracy was well under 1 mm - sampling & filling test tubes.
).