A Six-Axis Printed Arm With Real Reach

Nov 13, 2026 Last reply: 4 hours ago 4 Replies
Jump to replies (4)

Most hobby arms I see on forums are either tiny desktop toys or kits that cost as much as a printer. This one sits in a different category: a 3D-printed manipulator with six degrees of freedom, a 600 mm working radius, and a 2 kg payload rating. That is genuine reach for anything built mostly from printed parts, and it puts the project closer to light industrial gear than to the usual servo bracket builds.



The construction is where it gets interesting. The joints use cycloidal drives plus a split ring compound planetary gearbox, paired with Nema 17 steppers in 48 mm and 23 mm lengths and one 20 kg servo. Bearings are proper ball bearings in four sizes, with GT2 pulleys and belts, M4 and M5 threaded steel rods, and socket head screws from M2 up to M8. Feedback comes from magnetic encoders and endstop switches, driven by an Arduino, with motor drivers on heatsinked boards cooled by a 30x30x10 mm fan.



On the software side, it plays nicely with ROS, NVIDIA Isaac Sim, MATLAB and RoboDK, and there is a native platform covering Python, block coding and visual programming. An assembly manual exists as well, produced by assemblymaker.com, which is a real commitment most printed arm projects never make.



The open question is cost and feedback. Open-loop control runs about $400, while closed-loop is around $600. Also, the top speed and repeatability have not been stated anywhere, so that remains unknown for now.



Project by Arctos Robotics. Watch it on their YouTube channel:

formatting link


The choice of cycloidal drive joints in a printed build is the standout for me. Printing cycloidal reducers is fiddly and tolerance-sensitive, so going that route instead of simple gear trains seems like the reason this arm can hold a 600 mm radius without floppy links. The trade-off is likely print failure risk, since one bad layer in a cycloid disc can ruin the whole joint.

I am curious how the closed-loop version is set up. The magnetic encoders and endstop switches are listed, but is the $600 closed-loop option using those encoders on every joint, or only on some axes? And does the Arduino handle the encoder reading alone, or is there something else in the loop with it?

The maker has not said how the encoders are distributed across the joints, so I cannot answer that part. What is known is that the build incorporates magnetic encoders and endstop sensor switches, and that the system runs on an Arduino microcontroller board. The pricing difference is also on record: open-loop control comes in around $400 while closed-loop is roughly $600.

If anyone copies this, I would start with the closed-loop electronics from the beginning rather than upgrading later. Paying about $600 up front instead of $400 and then redoing wiring for encoders seems cheaper in total effort. I would also plan the cable paths before printing the arm sections, since the build already includes a dedicated cables path and retrofitting one is rarely clean.

Join the Discussion

Have something to add? Share your thoughts — no account required.

Didn't find your answer?

Ask the community — no account required