ROS Resources: Documentation | Support | Discussion Forum | Index | Service Status | ros @ Robotics Stack Exchange
Ask Your Question

sensor_msg:Imu format specification

asked 2017-03-01 04:22:14 -0500

Hi, I was confused by what data should go into the sensor_msg::Imu message. It has a field called linear_acceleration. The existing users and provider assume that it takes acceleration containing the gravity vector, and thus it doesn't actually correspond to a linear movement. I found this terminology surprising, expecting to find a gravity-subtracted acceleration. I also noticed that it doesn't appear to be documented whether orientation corresponds to a rotation from the body to some fixed frame or the opposite, and similar considerations for the other fields. A long discussion of the coordinate frame conventions can be found here!top...

My question is: is there any specification of the expected contents?


ps I ran into this when trying the ROS driver for my employer's LPMS IMUs which can be found here I made a blog post on my experience using it here:

edit retag flag offensive close merge delete


This appears to be a duplicate of #q255865. Could you please decide which one you'd like to keep and close / delete the other one?

gvdhoorn gravatar image gvdhoorn  ( 2017-03-01 04:43:25 -0500 )edit

Thanks. After writing the first attempt, I was led through the login / account creation process. When I finished this my post appeared to have vanished with not mention of moderation. Not knowing about this, I wrote it anew. Sorry about the extra work for you.

tobi_s gravatar image tobi_s  ( 2017-03-01 04:58:44 -0500 )edit

No problem. I just couldn't decide which one was more 'correct', so I thought I'd leave the choice to you.

gvdhoorn gravatar image gvdhoorn  ( 2017-03-01 05:41:18 -0500 )edit

2 Answers

Sort by ยป oldest newest most voted

answered 2017-03-01 05:49:58 -0500

gvdhoorn gravatar image

I think the current best / accepted practices are summarised in ros-infrastructure/rep#95, which is a PR that would add a REP (145 tentatively) entitled: Conventions for IMU Sensor Drivers.

I don't know what it's current status is though, you could perhaps post a comment on the PR.

edit flag offensive delete link more


Thanks, that is very useful! I added a comment, hoping that this would accelerate the path towards becoming official.

tobi_s gravatar image tobi_s  ( 2017-03-02 22:39:33 -0500 )edit

I've seen your comment. You might want to point back to your ROS Answers post, and provide a bit more context / background. I think input from a mfg of IMUs would be very welcome.

gvdhoorn gravatar image gvdhoorn  ( 2017-03-03 05:39:27 -0500 )edit

Thanks for the encouragement. I think what is described in REP-145 is perfectly reasonable. I'm not familiar enough with ROS to decide whether it's a good idea to post imu messages with potentially different frames (ENU vs NED) in the same topic.

tobi_s gravatar image tobi_s  ( 2017-03-05 20:10:04 -0500 )edit

The frame of the orientation quaternion will drift over time. W/out magnetometer, this will happen because of the lack of reference about the vertical axis, w/ magnetometer movement relative to nearby ferromagnets or electrical motors will affect the field, so the constant world frame is fictitious.

tobi_s gravatar image tobi_s  ( 2017-03-05 20:14:48 -0500 )edit

I'm not an expert, but I think in ROS we like to delegate those kind of things to other entities within your ROS node graph (ie: robot_localization fi).

gvdhoorn gravatar image gvdhoorn  ( 2017-03-06 00:22:32 -0500 )edit

Seeing this again: just to be clear, my previous two comments refer to two different issues, but the lack of space in the comments made me not write clearly enough.

Thanks for the pointer to robot_localization, will have a look.

tobi_s gravatar image tobi_s  ( 2017-03-06 03:42:53 -0500 )edit

answered 2017-03-06 21:46:04 -0500

This is not literally an answer to my question, but I think it is relevant enough to post it as an answer: there is a proposal to replace the imu_msg with something that divides more cleanly between the orientation and the IMU measurements:!msg...

In this proposal, the IMU estimates are also subdivided into accelerometer and gyroscope information.

edit flag offensive delete link more

Question Tools

1 follower


Asked: 2017-03-01 04:22:14 -0500

Seen: 1,314 times

Last updated: Mar 06 '17