Why Does map leaves lot of drift ? Is there any way to solve it

[saqihk]

Model: ugv_rover
After I ran the robot in one experiment, the map looked like a lot of drift, and after a minute, it changed the map and provided more drift.

How can we solve it ?

My second question is "my robot is connected to the wifi and I am running ssh commands through PC to control robot movements" If I want to experiment with different locations, then in that case, how will this work?

[mfruhner]

Hey,
are you still working on this and did you find a solution to the drift?
I am currently trying to run SLAM on my UGV 02 Rover and also see a lot of drift.
When I rotate in place, I can see in rviz2 that a half rotation is more than a full rotation in odom frame.
So i guess there is some miscalculation and therefore drift in SLAM and Nav2.

To your second question: What do you mean by different locations?
I am also controlling my robot over wifi and ssh. Once you have a map, you can save it and later load it for the location you are at.

Best regards,
Maik

[DUDULRX]

In what circumstances does drift occur, upload a video to see

[mfruhner]

Hey, sorry for the late reply.
Here is a video to show: https://youtu.be/ChnPqg9VOhQ
As you can see, the robot rotates too much in rviz. When I rotate back, it comes out at the starting position, so it is more or less consistent.

I am starting ugv_bringup.launch.py but without ekf or imu filter, as this makes it even worse.
I am using ros2 jazzy on a jeson orin nano.

Looking forward to your answer!
Best regards, Maik

[mfruhner]

Hey,
Did you have time to look at my video, yet?

[devashishraj]

@mfruhner @saqihk , how are you formatting data from IMU (ICM-20948 ) and source of data is chassis feedback or json interface ?

[devashishraj]

@DUDULRX @mfruhner @saqihk , does the code take account of magnetometer orientation ?

[DUDULRX]

嘿，你有时间看我的视频了吗？

Where does the odometer's posture originate?