• #56
• #57

Please pay more attention to the correctness of the model, when the robot has closed-loop chains like CURI.

• Forward kinematics
• Inverse kinematics
• Forward dynamics
• Inverse dynamics

When running the installation command

pip install -r requirements.txt

You may obtain this following error

  Running setup.py install for box2d-py ... error
  error: subprocess-exited-with-error
  
  × Running setup.py install for box2d-py did not run successfully.
  │ exit code: 1
  ╰─> [18 lines of output]
      Using setuptools (version 65.6.3).
      running install
      /home/clover/anaconda3/envs/rofunc/lib/python3.8/site-packages/setuptools/command/install.py:34: SetuptoolsDeprecationWarning: setup.py install is deprecated. Use build and pip and other standards-based tools.
        warnings.warn(
      running build
      running build_py
      creating build
      creating build/lib.linux-x86_64-cpython-38
      creating build/lib.linux-x86_64-cpython-38/Box2D
      copying library/Box2D/Box2D.py -> build/lib.linux-x86_64-cpython-38/Box2D
      copying library/Box2D/__init__.py -> build/lib.linux-x86_64-cpython-38/Box2D
      creating build/lib.linux-x86_64-cpython-38/Box2D/b2
      copying library/Box2D/b2/__init__.py -> build/lib.linux-x86_64-cpython-38/Box2D/b2
      running build_ext
      building 'Box2D._Box2D' extension
      swigging Box2D/Box2D.i to Box2D/Box2D_wrap.cpp
      swig -python -c++ -IBox2D -small -O -includeall -ignoremissing -w201 -globals b2Globals -outdir library/Box2D -keyword -w511 -D_SWIG_KWARGS -o Box2D/Box2D_wrap.cpp Box2D/Box2D.i
      error: command 'swig' failed: No such file or directory
      [end of output]
  
  note: This error originates from a subprocess, and is likely not a problem with pip.
error: legacy-install-failure

× Encountered error while trying to install package.
╰─> box2d-py

note: This is an issue with the package mentioned above, not pip.
hint: See above for output from the failure.

Thank you. I would like to ask what learning materials or papers are available for demonstration learning.of course,your code is very good information.

rf.zed.export: add a function that can set avi file output dictionary.

Will this package supports the isaaclab or isaac sim 4.0 ?

Require to finish the document of Xsens and Manus gloves

RuntimeError: Failed to acquire interface: omni::kit::IApp (pluginName: nullptr)
There was an error running python

Have already tried: https://forums.developer.nvidia.com/t/runtimeerror-failed-to-acquire-interface-omni-iviewport-pluginname-nullptr/240671/5

Ubuntu 20.04

作者您好，在成功复现rofunc相关功能后，近期开始使用rofunc框架对我们的数据进行处理。我们使用【Construct custom human model from Xsens data】中的代码对录制Xsens数据进行建模后显示报错KeyError: 'jL5S1' ,是否是因为所使用Xsens版本有细微的差别呢？想请问一下是否有对应的数据预处理解决方案？具体报错及isaacgym模块运行情况如下图所示，其中isaacgym界面会出现机器人的影子，但无法移动界面，点击相关按钮会出现提示【isaacgym无响应】。

• Standard gym-based RL environment using Isaac Gym
• Simple scenarios to train

• BaseMLPEncoder
• CNNEncoder
• ResnetEncoder
• ViTEncoder

• LQT
• LQT-bimanual
• LQT-control primitive
• LQT-feedback

Require to finish the document of behavior cloning from observation (BCO)

Dear authors,

Hello, thanks for your great work. I am working on the improvement of TP-GMM proposed by Sylvain Calinon. I need to implement it using Python and find the Rofunc package useful. I have several questions regarding the TP-GMM implementation.

1. I find you and Sylvain both implement TP-GMM using HMM, instead of GMM, which is not consistent with its definition and Matlab version. I know HMM definitely works here, but can you explain why to use it rather than GMM?
2. You use the marginal_model method to get the local GMM model in each frame. Looking at its definition, we can find the local model would have the same priors as the original model. But when the original HMM model is trained through EM, the priors are not updated and only the init_priors are updated. So I think the priors of local GMM models are not correct in this way.
3. In the reproduce method, why do you use LQR to estimate, instead of directly using GMR? The same question applies to the uni method of your TP-GMR implementation.

I initially referred to the Matlab version of TP-GMM provided by Sylvain. It is more detailed and consistent with the description in his papers. I do not understand why the Python implementation is a little different from Matlab. I find the Rofunc package keeps this difference due to the use of the pbd package. I am confused about why not just use GMM to implement TP-GMM. So I asked the above questions.

Thanks for your answers in advance.

After installed from source, here's the error:

(rofunc) ubuntu@ymh:~/Rofunc/rofunc/examples/simulator$ python curi_track_traj.py
Importing module 'gym_38' (/home/ubuntu/anaconda3/envs/rofunc/lib/python3.8/site-packages/isaacgym/_bindings/linux-x86_64/gym_38.so)
Setting GYM_USD_PLUG_INFO_PATH to /home/ubuntu/anaconda3/envs/rofunc/lib/python3.8/site-packages/isaacgym/_bindings/linux-x86_64/usd/plugInfo.json
Traceback (most recent call last):
  File "curi_track_traj.py", line 10, in <module>
    import rofunc as rf
  File "/home/ubuntu/Rofunc/rofunc/__init__.py", line 13, in <module>
    from .lfd import ml, dl, rl
  File "/home/ubuntu/Rofunc/rofunc/lfd/rl/__init__.py", line 3, in <module>
    from rofunc.lfd.rl.online import PPOAgent
  File "/home/ubuntu/Rofunc/rofunc/lfd/rl/online/__init__.py", line 1, in <module>
    from .ppo_agent import PPOAgent
  File "/home/ubuntu/Rofunc/rofunc/lfd/rl/online/ppo_agent.py", line 12, in <module>
    import torch
  File "/home/ubuntu/anaconda3/envs/rofunc/lib/python3.8/site-packages/torch/__init__.py", line 191, in <module>
    _load_global_deps()
  File "/home/ubuntu/anaconda3/envs/rofunc/lib/python3.8/site-packages/torch/__init__.py", line 153, in _load_global_deps
    ctypes.CDLL(lib_path, mode=ctypes.RTLD_GLOBAL)
  File "/home/ubuntu/anaconda3/envs/rofunc/lib/python3.8/ctypes/__init__.py", line 373, in __init__
    self._handle = _dlopen(self._name, mode)
OSError: /home/ubuntu/anaconda3/envs/rofunc/lib/python3.8/site-packages/torch/lib/../../nvidia/cublas/lib/libcublas.so.11: symbol cublasLtHSHMatmulAlgoInit version libcublasLt.so.11 not defined in file libcublasLt.so.11 with link time reference

I am making a comparison between learning using TPGMMBi and TPGMMRPCtrl as shown in Figure 3 of Birp's paper. I am using the data provided in the Rofunc example (example_tpgmmbi_RPCtrl.py) but I am having problems with the process of generating new trajectories using a model that has already been learned.

The learning process does correctly in both cases but when I call Representation.generate() with new data for the Task Parameterized the TPGMMRPCtrl is unable to function correctly. First of all I have made a modification because as the code was initially, the rest of the demonstrations always came out. This was because the data of lqr.x0_l, lqr.x0_r and lqr.x0_l inside of the function _bi_reproduce() were always filled by the values of self.repr_l.demos_xdx[show_demo_idx][0], self.repr_r.demos_xdx[show_demo_idx][0] and lself. repr_c.demos_xdx[show_demo_idx][0] respectively which made it use only the demos data as initial data (as it happens when using Repr.reproduce()). To solve this when calling reproduce I used the data stored in the task parameters as shown in the picture.

The value of lqr.x0_c has been also modified to use the self.repr_c.task_params["frame_origins"][show_demo_idx] value (IMPORTANT: here i dont need to use the [0] at the end due to the form of generate the value, but the structure is the same that in the other task values). This value has been generated using the code in the following image (lines 491-493), as it was seen that by default self.repr_c.demos_xdx[show_demo_idx][0] was used, which generated incorrect values as it uses the relationship of the demo data which changes when we want to use other parameters, therefore what is shown in the image generates results of the relationship between r and l depending on the task parameters data used in Repr.generate().

By making these modifications, it has been achieved that the results of the initial data (which are stored in the values of X0) are taken into account by the new trajectories using the learned models, but the end points of the tasks are not taken into account. My question is whether this is an incorrect step in applying RoFunc or whether it is an internal bug that has to be solved.
I attach visual results as well as the values of the trajectories where it can be seen that the initial point is taken into account but the final point is not.

Data numerically compared for the left trajectory: The result shows that the initial point is correctly taked into account but the final point no:

Data generated with the initial demos: Thst is generated by Repr.reproduce() and it is correct.

Data generated with new task parameters: Thst is generated by Repr.generate() and it is incorrect.

Finally, the part of the code where i have done the called for the Repr.generate()

Thank you very much for your help and sorry for the very long issue.

• operation space control
• joint space control

Write the URDF document using markdown, and save it in doc/source/simulator

Good morning, my name is Adrián Prados, PhD student at the University of Carlos III. I am currently working on imitation learning and I would like to use your framework to work with your datasets as well as to implement my own Imitation Learning algorithms. I have tried to install both versions, the stable and the nigthl version and both have given me problems. For the installation I have been following your steps .
There are problems in the initialisations of the different simulators, there are repeated functions that contradict each other (the init() function is defined several times in the same class generating problems) and also the function calls have changed with respect to the documentation. I was wondering if this is something temporary or if there is some kind of problem or bug, since I can start working with at least the stable version.

Additionally I would also like to comment that the documentation is quite incomplete and there is not much information on how to start working in more depth with this framework. I don't know if it's still under development or if it's a problem between versions or a bug.

In any case I would like to know if there is any way to solve it to be able to start working with this framework as it seems interesting, useful and relatively quick to use.

Online RL algorithms

• PPO
• SAC
• TD3
• A2C

Offline RL algorithms

• CQL
• Decision-Transformer
• EDAC
• TD3BC

Mixline RL algorithm

• AMP
• ASE
• Online Decision Transformer
• SoftGPT

For windows users , how do I install rofunc?

• Ant
• CURICabinet
• D4RL
• FrankaCabinet
• GymTasks
• Humanoid
• HumanoidAMP
• HumanoidASE

作者您好，近期在尝试复现rofunc的simulator部分，在安装isaacgym-1.0rc4-py3-none-any.whl文件时涉及到tensorflow的安装，报错【ERROR: pip's dependency resolver does not currently take into account all the packages that are installed. This behaviour is the source of the following dependency conflicts.
filelock 3.12.3 requires typing-extensions>=4.7.1; python_version < "3.11", but you have typing-extensions 4.5.0 which is incompatible.】显示typing-extensions 需要大于等于4.7.1；但是安装typing-extensions 4.7.1却会报错【ERROR: pip's dependency resolver does not currently take into account all the packages that are installed. This behaviour is the source of the following dependency conflicts.
tensorflow 2.13.0 requires typing-extensions<4.6.0,>=3.6.6, but you have typing-extensions 4.7.1 which is incompatible.】，显示又需要低版本typing-extensions。请问这种问题是否有相应的解决方案？还是我在isaacgym模块的安装上存在问题呢？非常感谢你能看到此处~

作者您好，最近希望借助【LQT with control primitives and DMP】中的代码对我们的数据进行处理，进行试验后发现这一节所采用的示例文件S.npy为三维numpy数组形式，但按照rofunc前期相应数据处理操作所得npy文件应为二维numpy数组形式，想请问如果需要运行【LQT with control primitives and DMP】应该对前期所得二维numpy数组形式的npy文件进行怎样的数据处理操作呢？自身在数据处理方面的知识较有欠缺，非常感谢您的项目带来的帮助~

After installed from source, if we run the example directly, there will be the following error:

ImportError: libpython3.8.so.1.0: cannot open shared object file: No such file or directory, 

Require to finish the document of Delsys EMG sensors

• iLQR
• iLQR-CoM
• iLQR-feedback
• iLQR-control primitive
• iLQR-dynamics
• iLQR-bimanual

我想使用您库中的带有障碍物的轨迹规划功能，但遇到了问题。

这个1.py就是

import rofunc as rf
from isaacgym import gymutil

args = gymutil.parse_arguments()
args.use_gpu_pipeline = False

frankasim = rf.sim.FrankaSim(args)
frankasim.show()

运行报错

我不知道我的安装对不对，我在conda 3.8中 ubuntu22.04，rofunc是python setup.py install安装的。
这些脚本执行成功了

sh ./scripts/install.sh
# [Option] Install with baseline RL frameworks (SKRL, RLlib, Stable Baselines3)
sh ./scripts/install_w_baselines.sh

可以帮助我吗？

• An easier-to-use simulator base class with cleaner logic, making it suitable for both constructing RL tasks and CPU-based planning.
• Update all robot sim

• GMR
• TP-GMM
• TP-GMR
• LQT
• LQT-Bi
• LQT-Hierarchical
• LQT-CP
• LQT-CP-DMP
• LQT-FB
• iLQR
• iLQR-COM
• iLQR-CP
• iLQR-Dyna
• iLQR-Obstacle

Human embodiment demonstration data contains human whole-body motion and motion of the object being manipulated

• Re-organize the datalab/poselib to fit the xsens suit, ensure that every joints are mapped properly
• Re-align the object motion data from Optitrack to human whole-body motion data in device/mmdoal
• Re-align the object motion with the human motion in the Isaac Gym simulator so that it conforms to the physical simulation, and provide the gap evaluation (the object motion should be the consequence of human manipulation)

• LQT
• LQT feedback / recursive form
• LQT with control primitive
• LQT with control primitive in a feedback / recursive form by using DMP

Require to finish the document of Optitrack

• ppo + state encoder
• sac + state encoder
• td3 + state encoder
• a2c + state encoder
• amp + state encoder
• ase + state encoder

RLlib is a powerful Reinforcement learning framework that provides plenty of baseline RL algorithm implementations.

• Standard gym-based RL environment using Isaac Gym
• Simple scenarios to train

When I imported curi.urdf into Isaac Sim and simulate, it started tweaking weirdly. How can I fix this?
https://www.youtube.com/watch?v=G0cjGfwl55Q