Skip to content

RSDK-14233: add KinematicModel to GetKinematics and FrameSystemConfig - #899

Merged
JohnN193 merged 9 commits into
viamrobotics:mainfrom
JohnN193:RSDK-14233-kinematic-model
Oct 13, 2026
Merged

JohnN193 merged 9 commits into
viamrobotics:mainfrom
JohnN193:RSDK-14233-kinematic-model

Conversation

@JohnN193

@JohnN193 JohnN193 commented Sep 18, 2026 •

Copy link
Copy Markdown
Member

Adds a typed KinematicModel message and carries it in GetKinematicsResponse and FrameSystemConfig. API half of RSDK-14233, tracked as RSDK-14747.

Changes

  • New messages in common.proto: KinematicModel, KinematicLink, LinkPose, Quaternion, EulerAngles, AxisAngles, JointType, KinematicJoint, JointLimits, MimicJoint, KinematicProperties.
  • GetKinematicsResponse.model and FrameSystemConfig.model. The existing bytes fields and the Struct are deprecated, not removed.
  • GetKinematicsRequest.include_collision_meshes and include_visual_meshes. Mesh bytes are only sent when requested, otherwise each mesh carries its path.
  • Mesh.original_file_source_path_diagnostic, the path a mesh was loaded from relative to the kinematics file.

Notes

  • One JointLimits per joint. A hardware and user split was rejected in the scope.
  • No generation field. Runtime change detection needs changes to the move RPCs and is separate work.
  • LinkPose.orientation is a oneof over the representations SVA files accept today, quaternion, orientation vector, Euler angles and axis angles. Settled with the scope, every existing file converts without loss and an author keeps the form they wrote.
  • No generated code is committed by hand. The ready-for-protos label has CI compile the protos, push the output to the branch and add protos-compiled.

Testing

buf lint, buf breaking against main, buf format and go build ./....

The rdk side starts with viamrobotics/rdk#6637, which builds against this branch. viamrobotics/rdk#6494 is the earlier prototype.

🤖 Generated with Claude Code

GetKinematics returns a format tag plus the bytes of a file, so every
consumer parses kinematics by hand and nothing checks their understanding
of the schema against ours. This adds a typed KinematicModel message and
carries it in both GetKinematicsResponse and FrameSystemConfig, which
today carry the same model as bytes and as an untyped Struct respectively.

The message reuses Vector3, Orientation, Geometry and Mesh. Link
orientation is a oneof over the representations SVA files already accept,
so an existing file converts without loss. Each joint carries hardware
limits and user limits separately so a module can validate one against
the other. Mesh gains source_path so a file can reference a mesh by
relative path and a response can identify a mesh whose bytes were left
out by the new exclude flags on the request.

The bytes fields and the Struct are deprecated rather than removed; the
server keeps filling them for older clients during the migration window.

Only the Go for common.proto and robot.proto is regenerated here, with
the pinned protoc-gen-go. Web output and proto docs still need the mise
tasks.

Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
JohnN193 and others added 4 commits October 12, 2026 10:13
Mesh bytes are opt in through include_collision_meshes and include_visual_meshes
instead of opt out. A joint carries one set of limits, the effective ones the
driver serves. JointType gains CONTINUOUS so URDF and the C++ SDK round trip.
KinematicProperties carries operating_freq_hz and max_linear_tcp_speed_mm_per_sec.
Mesh.source_path becomes original_file_source_path_diagnostic, since nothing
outside the driver resolves it. The generation field is removed, it needs move
RPC changes that are separate work. The name field is documented as set by the
SDK to the resource name.

Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
One line per field like the rest of common.proto. The design reasoning lives
in the scope and the PR, not in the proto.

Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
@JohnN193
JohnN193 marked this pull request as ready for review October 12, 2026 18:08
The ready-for-protos label has the compile protos workflow build and push the
generated output, so the hand run protoc-gen-go output comes back out.

Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
@JohnN193 JohnN193 added the ready-for-protos add this when you want protos to compile on every commit label Oct 12, 2026
@JohnN193 JohnN193 added question Further information is requested and removed question Further information is requested labels Oct 12, 2026

Copy link
Copy Markdown
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

If this is 1:1 vs the scope (which it looks to me like it is) then LGTM. If there were any meaningful divergences from the proto as proposed in the scope, could you please call that out with line comments so I can take a look?

@JohnN193

Copy link
Copy Markdown
Member Author

If this is 1:1 vs the scope (which it looks to me like it is) then LGTM. If there were any meaningful divergences from the proto as proposed in the scope, could you please call that out with line comments so I can take a look?

yeah its the same as whats in the scope!

JohnN193 and others added 2 commits October 13, 2026 11:00
The generated files take main's version so CI rebuilds them from the merged protos.

Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
@JohnN193 JohnN193 added question Further information is requested and removed question Further information is requested labels Oct 13, 2026
@JohnN193
JohnN193 merged commit bbe521a into viamrobotics:main Oct 13, 2026
6 of 7 checks passed
@JohnN193
JohnN193 deleted the RSDK-14233-kinematic-model branch October 13, 2026 17:44
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

ready-for-protos add this when you want protos to compile on every commit safe to test committer is a member of this org

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants