不能从 ONNX 元数据中获取“anchor_body_name”和“body_names”

[asxiaofengzi]

不能从 ONNX 元数据中获取“anchor_body_name”和“body_names”
[mujoco_sim-3] [ERROR] [1756288569.644669495] [walking_controller]: Caught exception in callback for transition 10
[mujoco_sim-3] [ERROR] [1756288569.644710105] [walking_controller]: Original error: OnnxPolicy: 'anchor_body_name' not found in model metadata. Please check the model.
[mujoco_sim-3] [WARN] [1756288569.644723703] [walking_controller]: Error occurred while doing error handling.
[mujoco_sim-3] [ERROR] [1756288569.644733142] [controller_manager]: After configuring, controller 'walking_controller' is in state 'unconfigured' , expected inactive.
[spawner-1] [ERROR] [1756288569.645262881] [spawner_state_estimator]: Failed to configure controller
[INFO] [spawner-2]: process has finished cleanly [pid 597]
[ERROR] [spawner-1]: process has died [pid 595, exit code 1, cmd '/opt/ros/humble/lib/controller_manager/spawner state_estimator walking_controller --param-file /tmp/motion_tracking_controller/temp_controllers.yaml --ros-args'].

[qiayuanl]

Can you try pulling the latest training code? I think this commit will fix the problem.

[Rookie629]

I've tried to add these two param manually with a python script and it works well(even Idk if it is the right way to do so lol). You can run this after you get the onnx from play.py. Btw I am bad at coding so this is fully create by Chatgpt:)

onnx_fixer.py