Segmentation fault when running joint_monkey.py on a remote Ubuntu Server

Hi there,
I have a remote Server with Ubuntu22.04 and NVIDIA Gefroce RTX 4090. I connect to the server through ssh in the Vscode or MobaXterm. When I run python joint_monkey.py, it always turn to Segmentation Fault, which confused me for a long time. I noticed some issues that there might be some problem while using a headless server, but my server is with a monitor. When I try to set the graphics_device_id to -1, I will get another error: Failed to create viewer.
Besides, I can run the RL examples succesfully with the headless mode. For example, python train.py task=Ant headless=True. It’s always OK. Now I wonder whether there is something I can do to get the display window. I have seen that someone succeed but I don’t know what I should do now. Thank you for any help!

Best wishes,
Peter

Here is some output:
python joint_monkey.py
Importing module ‘gym_37’ (/home/zhangtianyi/Robot2/isaacgym/python/isaacgym/_bindings/linux-x86_64/gym_37.so)
Setting GYM_USD_PLUG_INFO_PATH to /home/zhangtianyi/Robot2/isaacgym/python/isaacgym/_bindings/linux-x86_64/usd/plugInfo.json
WARNING: Forcing CPU pipeline.
Not connected to PVD
+++ Using GPU PhysX
Physics Engine: PhysX
Physics Device: cuda:0
GPU Pipeline: disabled
Segmentation fault (core dumped)