Six axis camera arm fighting jitter

Oct 22, 2026 Last reply: 1 day ago 2 Replies
Jump to replies (2)

Stop motion photography punishes any movement at all, and that is exactly the problem this build runs into. The maker, a freelance filmmaker, has put together a 6 DOF motion control arm to hold a mirrorless camera with a wide lens, and the constant headache is vibration or shake whenever the shoulder pitch and elbow pitch joints move together under load.



The hardware itself looks well thought out. The arm is built from black custom structural parts printed in PETG-CF and PAHT-CF, and it carries a set of Robstride actuators, one per joint: an EduLite 05 with 6 N·m on the shoulder pan, a pair of RS10P units with 42 N·m each on shoulder and elbow pitch, RS00 units with 14 N·m on roll and pitch, and an RS05 with 5.5 N·m on the final pan. The whole bill of parts and print material came to roughly 1k USD, which seems very reasonable for this kind of rig.



The control side is where the interesting work happens. A PC builds trajectories by polynomial interpolation between pose keyframes, smoothed with a quintic curve. An STM32 running in MIT mode plays the trajectory back on its own clock, sampling the curves at 400 Hz into MIT commands every 2.5ms. The controller layers in saved gravity and friction estimates, estimated inertia times acceleration, enabled ripple corrections, and an integral term accumulated from position error over time.



There has been real progress: major stepping issues on some actuators were fixed with differential gear calibration, and adding the integral component helped a little. The shake is not fully solved though, and the maker is still weighing a lower I action on the position loop plus local PID control per joint. The plan to pair it with a GoPro ILS and follow focus makes it worth watching.


This is a nice build. One thing I want to understand better: with the PC generating the trajectory and the STM32 playing it back, how does the timing stay locked between the two? I would worry about the arm drifting if the two sides are not sharing a clock, especially over long stop motion moves.

The maker has actually covered that part. The STM32 plays the trajectory using its own clock rather than trusting the PC for timing, and it samples the curves at 400 Hz, turning them into MIT commands at 2.5ms intervals. So the microcontroller is the timekeeper and the PC only hands over the interpolated trajectory beforehand.

Join the Discussion

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

Didn't find your answer?

Ask the community — no account required