Timeline for EKF filter node not working correcty
Current License: CC BY-SA 4.0
7 events
| when toggle format | what | by | license | comment | |
|---|---|---|---|---|---|
| Mar 22 at 13:48 | comment | added | automatom | I address wheel slip in the answer to the question above. The filter won't solve that for you. | |
| Mar 22 at 13:36 | comment | added | Jesús | Hi again. I did what you told me: I made the odom_publisher only publish linear.x velocities (not poses or orientation) and I achieved to "mix" the rotation speed read by the IMU (gyro) with the linear from the odom. The question is, is there any way using robot_localization pkg who allows me to detect linear slip? When my robot rotates while slipping (in the same point) all goes good, the problem is correcting linear slips, as the node takes the linear.x from odom and ignores x_accel or x_speed from IMU | |
| Mar 21 at 8:50 | comment | added | automatom |
A nav_msgs/Odometry message contains pose and velocity (twist) data. You can fuse whatever variables you want from that message; that's what the odom0_config is for. So if you set the pose variables to false and the velocity variables to true, it will fuse those instead. If you only fuse velocity data, it will get integrated into pose data. Your covariance will slowly grow (and without bound), but that won't matter.
|
|
| Mar 21 at 8:49 | history | edited | automatom | CC BY-SA 4.0 |
added 38 characters in body
|
| Mar 20 at 14:20 | comment | added | Jesús | Wouldn't be enough with just simply setting to false the x_pos and y_pos values from the odom0_config? Does it matter if my odom_publisher's still publishing both pose and twist when I have set those variables to false? Which source (odom or imu) is going to tell x_pos and y_pos? Thx | |
| Mar 20 at 14:10 | comment | added | Jesús | Thank you for your response!! I added an example of the two messages (odom and IMU) that I have when the robot is moving | |
| Mar 20 at 11:33 | history | answered | automatom | CC BY-SA 4.0 |