Allow a subset of <geometry> instead of <inertia>

[osrf-migration]

Original report (archived issue) by Shane Loretz (Bitbucket: Shane Loretz, GitHub: sloretz).

---

Issue

The <inertial> element requires an inertial matrix <inertia>. When creating models I often find myself plugging the geometry of the <collision> into one of the formulas on this wikipedia page, which can be prone to errors as I often end up changing the geometry.

Proposal

The proposal is to allow a subset of the shapes in <geometry> (box, cylinder, sphere) as an alternative to <inertia>. The inertia matrix values would be calculated by assuming uniform density.

<link name="some_link">
  <collision name="my_collision">
    <geometry>
      <sphere>
        <radius>0.5</radius>
      </sphere>
    </geometry>
  </collision>
  <inertial>
    <mass>1.0</mass>
    <sphere>
      <radius>0.5</radius>
    </sphere>
  </inertial>
</link>

Related

Issue #95 combined with this feature would make it much easier to specify the inertia on a link with multiple <collision>.

[osrf-migration]

Original comment by Ian Chen (Bitbucket: Ian Chen, GitHub: iche033).

---

• changed title from "Allow a subset of <geometry> instead of <interia>" to "Allow a subset of <geometry> instead of <inertia>"

[osrf-migration]

Original comment by Ian Chen (Bitbucket: Ian Chen, GitHub: iche033).

---

I remember there was talk about changing the default inertia value (from identity matrix) to be something computed on geometry type, mass, and size. So if the <inertial> sdf element is not present then compute appropriate values potentially using ignition::math::MassMatrix3d

[osrf-migration]

Original comment by Kunal Tyagi (Bitbucket: kunaltyagi).

---

Has there been a progress on either issue? I would like to contribute to this coz it is a regular problematic issue for my team.

[scpeters]

we also discussed this a bit in #95 (comment)