May 01, 2020
I'd like to design my own robotic arm. I'd like it to be roughly twice the size of AR2/3, approx. 1.2 m. I'd like to have accuracy of a few tenths of a millimeter. When I calculated what angle it corresponds to, it turned out I'd need gearboxes with <0.1 arcmin backlash. Even if obtainable, they'd cost a fortune. How come, then, AR2/3 robot has so good accuracy using cheap hobbyist parts? Why doesn't it go "flapping around in the breeze, wiggle wiggle wiggle yeah", as some might say? Am I missing something? I understand it's not such a big problem when parts are under constant tension in the same direction. In that case the backlash doesn't come into play... But what about rotating the arm around horizontal axis so it passes the vertical? Wouldn't the end effector just fall freely a few millimeters (based on a few arcmin backlash) when the gravity starts pulling in the other direction?
Oct 15, 2018
Noob questions warning. I want to make a welding robot arm for our family company and I've been looking for inspiration. I found your video and github page, and it's just pure GOLD. Thanks for all this. I'l be buying the CADs at the very least. 1) Do you think this robot will be enough for welding? 2) I want to teach it its moves by example. I don't know how exactly I'll be doing it, but maybe the simplest method would be making a second identical arm with encoders instead of motors. Or use a SpaceMouse as a force control for teaching. I don't know yet. I'll figure things out in due time, after I first played with the robot as it is. But in general, do you think it's feasible? (I have no idea about any of this yet, but I'm generally quite good with computers, technology, and figuring things out) 3) Aren't servos better than stepper motors (something about losing steps and knowing current position)? If I wanted to use servos, would it be possible without major changes? Thanks again for making this project! Andrzej.