Rviz and real-world movement differs


I have Stretch RE2, on ROS1 Noetic. When I tried the gmapping (roslaunch stretch_navigation mapping.launch), I stumbled on the issue of the opposite movement of rviz and real-world. Basically, when I tried to teleop forward, it will show backward in rviz and vice-versa. I tried to tf view_frames (attached) and I don’t see anything wrong. However, when I tried to add IMU_mobile_base, it showed and error: For frame [imu_mobile_base]: Frame [imu_mobile_base] does not exist. I tried to echo it, it showed some of numbers however i’m not sure if that’s how it should work. I suspect IMU didn’t get tf, thus it gave this opposite movement between rviz and real-world. Other than that, I couldn’t fix the gmapping, the scan laser keeps changing while the robot moves. Perhaps if I could fix the opposite movement issue I will also solve the gmapping issue? Please advise. Thank you in advance.

Below I upload the video of the demonstration. It will be appreciated if anyone can give advise on this. Thank you.

And for the laser scan that somehow moves when the stretch robot moves:

Hi @Mustofa, I suggest we start debugging this at a lower level in the software stack to figure out where the issue stems from. Stretch’s ROS packages build on a low level Python library called Stretch Body. There’s a command line tool that lets us teleop base movements directly at the Stretch Body level. Would you run the following tool in a terminal:


You’ll a menu like this:

1: rate slow
2: rate default
3: rate fast
4: rate max
w: CW/CCW 90 deg
x: forward-> back 0.5m
y: spin at 22.5deg/s

f / b / l / r : small forward / back / left / right
F / B / L / R : large forward / back / left / right
o: freewheel
p: pretty print
q: quit

Try pressing f a few times. The robot should move forward. Additionally, the tool will print out the robot’s odometry. It’ll look like:

X (m) 0.03756735929715982
Y (m) -4.779596338266856e-06
Theta (rad) 6.283023855446416
X_vel (m/s) 0.001202183843753624
Y_vel (m/s) 0.0
Theta_vel (rad/s) 6.977221664866586e-07
Pose time (s) 36.72699999809265
Timestamp PC (s): 1702765346.9285815

Positive X axis is the forward direction for the base, so you should see X (m) steadily increase as you press f.

Hi Binit Shah! Thanks for the cooperation. I tried the stretch_base_jog.py, and I observed that X (m) steadily increase as I pressed f. However, I think the robot was moving toward the negative X axis (backward, as shown in the video). It will be appreciated if you could provide me with the documentation about axis coordinate of the robot (currently I couldn’t find one in the doc., the most related one I found: hardware_guide_re2

Hi @Mustofa, this behavior is unusual. If you’re still experiencing this issue, would you reach out at support@hello-robot.com?

Also, this is what the mobile base coordinate frame looks like:


The x axis is aligned with the robot moving forward and the y axis point to the robot’s left. This forms the robot’s right-handed base_link frame.