Hi

I am planning on building this robot, and for the documentation purposes, I will have to calculate its repeatability.

The AR4 MK3 User manual says ±0.2mm, and I emailed Chris about this. He said this was experimental measured over several trials, and I believe this is the right way to test it finally. However, for my documentation, I will need to calculate it manually.

I did try calculating it and got 0.55mm repeatability, without even considering the gearbox play.

I calculated this by calculating the Angle per step for each axis motor, considering the microsteps and gear ratio. So for example, for the first axis, I got 0.02250 degrees per step. Then considering the robot in the fully extended position, the length was 628mm. When the axis 1 motor moves 1 step, it rotates the arm by 0.02250 degrees. This results in a 0.2466mm movement in the end effector.

Similarly I calculated for all remaining axis and summed this values. Basically this value represents the resolution of the robot, a.k.a the least count. The smallest distance that the end effector can move, when moved by Axis 1. By this approach, I got 0.55mm repeatability.

I am not sure if this approach was correct. Can someone help me with the correct method / approach to calculating the repeatability of the robot? Also if someone has done it already, it would be great if you guide me on the same.

Thanks in advance

Realistically, I think the most robust way to get the actual accuracy of the robot is probably to gather data from the real thing. There's a lot that you can't realistically account for in terms of play in joints and bearings and mechanical backlash and what not. I do respect the hustle of doing those calculations. I'll let someone smarter and more interested in the math take over for that one. :P