I backed a scara robot on kickstarter (
http://www.fluxintegration.com/ally) thinking that I would use it for PNP. I ultimately got tired of waiting for them to deliver and took the plunge in creating my own. Fast forward a year later and I'm still anxiously awaiting delivery of my scara, they are saying they will ship before Christmas (but they've also been saying this for a year now). If I do end up seeing it I will definitely be trying to make it talk to openpnp, but I am glad I built my own. We've turned out hundreds of flawless boards while we've waited for these guys to perfect the ALLY.
I posted the same link in the flux integration forum last week and got some interesting feedback from the developer of the ALLY scara arm I am waiting on, about the DotBot he said:
"They are using harmonic drives which should always cause concern when offered on a low cost machine. Harmonic drives need to be machined to incredible tolerances to have the precision and stiffness required for any sort of robot arm. That is why they cost > $500 per unit for average quality units. You are paying for the machining tolerances and the materials science of the components so they do not wear. If you look at the data sheets for low quality harmonic drives you quickly see that they are not magic. They have significant lost motion when torque is applied and the overall accuracy over 360 degrees is horrible when compared to those used for industrial robot arms. Robot arms are hard to build because any error in the joint is amplified by the reach of the robot arm. Counting on the same mechanical components that industrial robot arms use just machined cheaply and inaccurately gives you a cheap inaccurate robot arm not a precision low-cost robot arm.
I did want to provide a simple example of a repeatability calculation for a typical SCARA robot arm. This is using the specifications for an industrial grade harmonic drive from a U.S. company at the higher-end of quality, http://nacdrivesystems.com/technical.html. They state that there industrial grade harmonic drive has +/- 3 arc/min or +/- 0.05 degrees of backlash (lost motion) at no load, so no torque applied. If the robot arm has a 16" [400 mm] reach the major component of the repeatability calculation for the shoulder and elbow would be 16 * sin(0.05) + 8 * sin(0.05) = 0.021" [0.531 mm]. This is almost 27 times the claim for repeatability that Dobot makes."