Why no imu_calib for linorobot2?

[pitosalas]

I am having a little trouble with my new linorobot2 build. When stationary it shows in rviz2 as if it is slowly rotating. If I disable imu in ekf.yaml then the drifting stops. So I thought about imu_calib. Is there a technical reason why you don't recommend that in linorobot2?

[PaulBouchier]

No technical reason. However, at this point in time, I have to recommend hippo5329's fork. It has better documentation and better test utilities. I'm slowly working toward updating the upstream repo but Thomas Chou's hippo5329 is better in most respects at this time.

That said, he doesn't have a utility called imu_calib - is that something you have seen elsewhere?

IMU's are known to drift, so you might get it to read 0 rotation but over time, it will drift again. Do you have odometry flowing properly? Odometry and IMU get fused in the EKF and odometry should stop a drifting IMU from causing spinning.

[pitosalas]

Hi Paul Thanks. I am using hippo5329’s fork but I track the original too. I had used Linorobot(1) for a long time before I went to linorobot2. In (1) there was a imu_calib plus instructions. I can’t find (1) anymore he must have hidden it. Here’s one but I am not positive it’s the one that (1) used: https://index.ros.org/r/imu_calib/

IMU's are known to drift, so you might get it to read 0 rotation but over time, it will drift again. Do you have odometry flowing properly? Odometry and IMU get fused in the EKF and odometry should stop a drifting IMU from causing spinning.

Yes. You’re on the Homebrew Robitcs list so you’ve seen my posts there. It's do have spinning but oddly sometimes when I power up the robot it does not spin and sometimes it does. I can’t pin down the pattern. One thing that I want to check is whether I innocently am handling the robot at the moment I power it up because I think the IMU goes through a self-calibration process during the first 8 seconds. Do you know about that? When I look at the rotation coming out of the IMU, and it is totally still, what I saw yesterday is a constant zero yaw interspersed every so often (3-5 seconds) by a very short pulse reading of rotation and then back to zero. Not sure yet whether that’s normal, imagined, spurious or what. A question I have is whether this IMU (and newer ones) do this self calibration and so the old imu_calib process is actually no longer needed. Best, Pito Boston Robot Hackers && Comp. Sci Faculty, Brandeis University (Emeritus)

…

On Jan 17, 2026, at 10:07 AM, Paul Bouchier ***@***.***> wrote: PaulBouchier left a comment (linorobot/linorobot2_hardware#112) No technical reason. However, at this point in time, I have to recommend hippo5329's fork. It has better documentation and better test utilities. I'm slowly working toward updating the upstream repo but Thomas Chou's hippo5329 is better in most respects at this time. That said, he doesn't have a utility called imu_calib - is that something you have seen elsewhere? IMU's are known to drift, so you might get it to read 0 rotation but over time, it will drift again. Do you have odometry flowing properly? Odometry and IMU get fused in the EKF and odometry should stop a drifting IMU from causing spinning. — Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you authored the thread.Message ID: ***@***.***>

[PaulBouchier]

Good to know you have such a long history with this Pito.

Here's the imu_calib repo: https://github.com/linorobot/imu_calib
You'll find the rest of the linorobot 1 repos there, under the linorobot user.

I'm sorry but there's too much traffic on HBRC list for me to monitor.

Whether the imu self calibrates on startup is imu-dependent. The bno055 and 085 do, and if the robot is moving at that time the calibration may be wrong. Also, you are advised not to use the magnetometer because it can inject spurious readings, according to some DPRG guys. Which imu are you using?

Steady zero reading with occasional spikes sounds like a sensor or electrical noise issue.