Using a Gimbal with ZED-X Camera — Is It Supported and Recommended?

Hardware ZED X
1

Hello everyone,

I recently purchased a ZED-X stereo camera and I am exploring options for mechanical stabilization.
My main question is whether a gimbal can be used effectively with the ZED-X, and if so, under what constraints.

More specifically, I would like to understand:

  • Is the ZED-X designed to be mounted on a gimbal, or is it expected to be rigidly mounted?

  • Are there known issues when using a gimbal with a stereo camera (e.g. rolling shutter effects, baseline distortion, synchronization, calibration drift)?

  • Does Stereolabs provide any official recommendations or best practices regarding gimbal usage?

  • Are there specific gimbal models or stabilization approaches (2-axis vs 3-axis, IMU-assisted, software stabilization) that work better with ZED-X?

  • How does gimbal motion interact with depth estimation, visual odometry, and SLAM in the ZED SDK?

My use case involves mobile / UAV-mounted perception, so vibration isolation and motion smoothness are important, but I want to avoid degrading stereo depth quality or breaking calibration assumptions.

If anyone has practical experience using ZED-X (or similar stereo rigs) on a gimbal, I’d really appreciate insights, pitfalls, or example setups.

Thanks in advance!

2

Hi @SergeyMalashenko
Welcome to the Stereolabs community.

The ZED X camera has been designed targeting robotic applications, so it can work even if installed on a gimbal.

The ZED X has 2 Global Shutter CMOS sensors, so it’s not affected by this kinf of issues.

We do not have this information. Maybe other customers can share their experience.

A gimbal can compensate for camera self-rotation; the SLAM algorithm of the ZED SDK estimates camera ego-motion by using visual and inertial information. As long as the camera’s overall movements remain consistent with those of a robot, the SLAM results are unaffected.

Just an example, if a robot is rotating clockwise at 1 rad/sec and the gimbal compensates this movement, rotating the camera counterclockwise at 1 rad/sec, the SLAM algorithm will report 0 rad/sec as rotation speed, “thinking” that the robot is steady.

In this case, a good gimbal can only improve the final results, not degrade them.