Hacktoberfest 2026：维护者为十月标记出来的 issue，仍然开放、适合新手。 浏览 Hacktoberfest issue

Attachment frames's inertial property

未关闭
#477 0 条评论 0 个 reaction 已指派 0 人 在 GitHub 查看

还没有人认领这个 Issue。

评估

难度
3/5
预计耗时
1-2 天
新手友好度
35/100
Issue 类型
缺陷
描述清晰度
基本清楚
活跃度
停滞
技术栈
python
领域
tooling

调研方向

从 dm_control/mjcf/README.md 的 attachment-frames 部分开始，并使用其中展示的 Camera、parent-body 和 attach 代码复现该问题。将生成的 MJCF 与所要求的输出进行比较；当附加的 body 保留其 inertial 元素，且未引入额外的 unnamed_model body 时，即表示完成。

由索引模型根据 Issue 内容生成。

描述

Hello,

I am trying to attach a body to another one, similar to your example here of child and parent.

However, my child body does have inertial-properties. Attaching the following child body

class Camera:
    #===================#
    #  C O N S T A N T  #
    #===================#
    _com_pos = (0, 0, 0.015)
    _dclass = 'wam\\viz'
    _mass = 0.095
    _mesh_name = 'intel_realsense_l515'
    _name = '\\rgb'
    _quat = (0, 0, 1, 0)
    #===============================#
    #  I N I T I A L I Z A T I O N  #
    #===============================#
    def __init__(self, name, fovy):
        # Make a MujoCo Root
        self.mjcf_model = mjcf.RootElement()
        # Add a body element: No need to add position info, as that will be added based on which `site` this body is going to be attached (https://github.com/google-deepmind/dm_control/blob/main/dm_control/mjcf/README.md#attachment-frames)
        self.camera_body = self.mjcf_model.worldbody.add('body',
                                                         name = name)
        # Add Inertial properties
        self.camera_body.add('inertial',
                             pos = self._com_pos, 
                             mass = self._mass)
        # Add the Geometry (https://github.com/google-deepmind/dm_control/blob/ed424509c0a0e8ddf7a43824924de483026ad9cc/dm_control/locomotion/soccer/humanoid.py#L50)
        self.camera_body.add('geom',
                             mesh = self._mesh_name,
                             dclass = self._dclass)
        # Add camera sensor
        self.camera_body.add('camera',
                             name = name+self._name,
                             pos = (0, 0, 0),
                             quat = self._quat,
                             fovy = fovy)

as such in the parent body

        # Add camera locations
        # An empty array to hold camera-sites
        camera_site = []
        for x in range(len(self._camera_pos)):
            camera_site.append(self.summit_body.add('site',
                                                    pos = self._camera_pos[x],
                                                    quat = self._camera_quat[x],
                                                    dclass = self._site_name))
        # Create camera
        # Initialize empty list of camera
        cameras = []
        for x in range(len(self._camera_pos)):
            cameras.append(Camera(name = self._name_summit+'_'+prefix+self._camera_name[x],
                                  fovy = self._camera_fovy[x]))
        # Attach the cameras
        for x in range(len(self._camera_pos)):
            camera_site[x].attach(cameras[x].mjcf_model)

causes the generated file to introduce an unnecessary body called unnamed_model, which is derived from the name of the worldbody (which, in this case, was initialized without a name).

<body name="smt_0" pos="0 0 0">
      <inertial pos="0 0 0.37" mass="125" diaginertia="1.391 6.8529999999999998 6.125"/>
      <geom name="smt_0\viz\base_link" class="summit\body\viz" mesh="base_link_summit"/>
      <body pos="0 0.36199999999999999 0.373" quat="0 0 0.70710700000000004 0.70710700000000004" name="unnamed_model/">
        <body name="unnamed_model/smt_0\front\camera\intel">
          <inertial pos="0 0 0.014999999999999999" mass="0.095000000000000001"/>
          <geom name="unnamed_model//unnamed_geom_0" class="unnamed_model/wam\viz" mesh="unnamed_model/intel_realsense_l515"/>
          <camera name="unnamed_model/smt_0\front\camera\intel\rgb" class="unnamed_model/" fovy="55" pos="0 0 0" quat="0 0 1 0"/>
        </body>
      </body>
</body>

What I am aiming for is a representation similar to your example here, such that I should have:

<body name="smt_0" pos="0 0 0">
      <inertial pos="0 0 0.37" mass="125" diaginertia="1.391 6.8529999999999998 6.125"/>
      <geom name="smt_0\viz\base_link" class="summit\body\viz" mesh="base_link_summit"/>
      <body pos="0 0.36199999999999999 0.373" quat="0 0 0.70710700000000004 0.70710700000000004" name="unnamed_model/">
          <inertial pos="0 0 0.014999999999999999" mass="0.095000000000000001"/>
          <geom name="unnamed_model//unnamed_geom_0" class="unnamed_model/wam\viz" mesh="unnamed_model/intel_realsense_l515"/>
          <camera name="unnamed_model/smt_0\front\camera\intel\rgb" class="unnamed_model/" fovy="55" pos="0 0 0" quat="0 0 1 0"/>
      </body>
</body>

but I am not allowed to do this because AttributeError: <inertial> is not a valid child of <worldbody>.

Question;
How can I get such a clean representation while introducing inertial properties for attached-bodies?

主要语言
Python
星标
4.7k
派生
765
30 天内没有已合并 PR

贡献指南

打开贡献指南

从这里开始

  1. 先读完整个 Issue，再读项目的贡献指南。
  2. 在 Issue 下留言说明你要接手 —— 这能避免两个人做同样的事。
  3. Fork 仓库，在一个分支上完成修改。
  4. 提交 Pull Request，并在描述里引用这个 Issue 编号。

google-deepmind/dm_control 的其他 Issue

查看 google-deepmind/dm_control 的全部 Issue

相似的 Issue

更多 Python Issue

把新 issue 发到你的邮箱

精选适合新手参与的 GitHub issue 摘要。