ROS Resources: Documentation | Support | Discussion Forum | Index | Service Status | ros @ Robotics Stack Exchange
Ask Your Question
0

robot_localization navsat_transform yaw offset and phidgets imu

asked 2017-10-25 05:39:59 -0500

blasalat gravatar image

Hi, I am currently using the robot_localization package for the state estimation of my robot. I also have a RTK GPS Device which I would like to include in robot_localization. The problem is that I don't know my yaw offset because I am using a phidgets IMU and I am not including the magnetic field when using the madgwick orientation filter because that causes poor results as the IMU is located close to the motor.

My idea would be to calculate the yaw offset by getting the course over ground from my GPS-Receiver. I would then first start navsat_transform when I calculated the new yaw offset or I would add a service to navsat_transform to set the yaw offset during runtime.

Has any one had similar problems and a good solution?

edit retag flag offensive close merge delete

1 Answer

Sort by ยป oldest newest most voted
1

answered 2018-01-24 04:00:30 -0500

Tom Moore gravatar image

It's not so much that you need the offset; you just need an earth-referenced heading. The yaw_offset and magnetic_declination parameters are there to allow you to compensate for the fact that (a) most magnetometers read 0 at north, not east, and (b) they read 0 at magnetic north, not true north.

So you could just calculate that heading from course over ground and generate a sensor_msgs/Imu message that navsat_transform_node would listen to instead of your actual IMU message. Just make sure it conforms to standards: a heading of 0 means you are moving directly east, and turning left is positive rotation. I would imagine course over ground assumes 0 at north, so you can either manually adjust for that in your course-over-ground-heading-node, or you can set yaw_offset to pi/2.

edit flag offensive delete link more

Question Tools

3 followers

Stats

Asked: 2017-10-25 05:39:59 -0500

Seen: 1,122 times

Last updated: Jan 24 '18