Intrinsic Inference
Based on published sources · Not tested by us ·
Overview
Preview framework for managing model-serving lifecycles and exposing inference services to ROS 2. [S001]
Intrinsic Inference manages model staging, loaded-model state and request forwarding to an inference backend. Its ROS node exposes that machinery through services. It does not supply a trained robotics policy just by installing the framework, and the repository explicitly labels the current work a preview that is not stable. [S001] [S003] [S004]
Specifications
| Detail | Description |
|---|---|
| Intended users | ROS 2 developers integrating an inference backend Editorial assessment [S001] · Checked 2026-09-30 |
| Required software | Bazel; NVIDIA Triton server; quickstart uses Docker image 26.07-py3; CycloneDDS for the documented ROS client/server route; A separately provided model repository Reported by source [S003] · Checked 2026-09-30 |
| Installation | Prepare the example model repository, start Triton in explicit model-control mode, then launch the Bazel InferenceNode target and make its custom interfaces visible to the ROS client. Reported by source [S003] · Checked 2026-09-30 |
| Uses | Stage and reconcile model assets; Proxy inference through a backend; Query backend readiness and model metadata from ROS Reported by source [S003] [S004] · Checked 2026-09-30 |
| File formats | ONNX in the demo; Protocol Buffers model manifests; ROS service definitions Reported by source [S003] [S004] · Checked 2026-09-30 |
| Development status | prototype Source-researched component with no release or installation validation by this publication. Editorial assessment [S001] · Checked 2026-09-30 |
3D preview
A robot-part preview is not applicable to this software project profile.
Installation and use
Start the backend before the ROS wrapper
The ROS quickstart prepares an external DenseNet ONNX model directory and starts Triton using the 26.07-py3 container image. Triton must run with --model-control-mode=explicit, because the framework manages model loading and unloading itself. It then launches inference_node_main with the model repository path and Triton's gRPC address. [S003]
For an initial check, make the custom inference interfaces available in the client workspace, source that workspace and call the documented server_live service. Querying model metadata is the next check before attempting useful inference. The guide selects CycloneDDS for both the Bazel-built node and its external ROS client; do not assume the Zenoh settings from Core's visualization example apply here. [S003]
Serving code is separate from the selected model
The core reconciles desired and actual model state, stages files and proxies requests. Compute demand therefore depends on the chosen model and backend; this review establishes no universal GPU, memory or throughput figure. The software repository carries Apache-2.0, but that does not resolve the rights of the separately downloaded model. [S002] [S003] [S004]
This profile pins an unreleased commit snapshot. We have not started Triton, downloaded the example weights or measured inference through the ROS wrapper. [S090]
Source files and licenses
Useful open software component with documented development workflow; dependency rights, configuration matching and runtime results remain separate.
| Layer | Availability / license |
|---|---|
| software | available · Apache-2.0 Inspected root license for this repository. External dependencies, vendor SDKs, models and third-party geometry are not relicensed by it. [S001] [S002] |
| model weights | available · License not fully assessed The quickstart downloads a separate DenseNet ONNX model. Exact model licensing and workload qualification are not established by this software review. [S003] |
| documentation | available · Apache-2.0 Inspected root license for this repository. External dependencies, vendor SDKs, models and third-party geometry are not relicensed by it. [S001] [S002] |
No particular model weights or dataset is offered or assessed by this component profile.
References
Links identify the documentation and revisions used in this profile. Access dates record when each source was retrieved.
- S001 · Intrinsic Innovation LLC / project contributors
intrinsic-inference: README.md
Accessed 2026-09-30 · 7760abb8097484a8168e34848403ca4b165e4480 / README.mdSource notes
Pinned source inspected for the specific workflow, requirement or license claims cited.
- S002 · Intrinsic Innovation LLC / project contributors
intrinsic-inference: LICENSE
Accessed 2026-09-30 · 7760abb8097484a8168e34848403ca4b165e4480 / LICENSESource notes
Pinned source inspected for the specific workflow, requirement or license claims cited.
- S003 · Intrinsic Innovation LLC / project contributors
intrinsic-inference: intrinsic_inference/ros/inference_node/README.md
Accessed 2026-09-30 · 7760abb8097484a8168e34848403ca4b165e4480 / intrinsic_inference/ros/inference_node/README.mdSource notes
Pinned source inspected for the specific workflow, requirement or license claims cited.
- S004 · Intrinsic Innovation LLC / project contributors
intrinsic-inference: intrinsic_inference/core/README.md
Accessed 2026-09-30 · 7760abb8097484a8168e34848403ca4b165e4480 / intrinsic_inference/core/README.mdSource notes
Pinned source inspected for the specific workflow, requirement or license claims cited.
- S090 · Intrinsic Innovation LLC / GitHub metadata
Pinned repository metadata
Accessed 2026-09-30 · 7760abb8097484a8168e34848403ca4b165e4480Source notes
Immutable commit snapshot; no published stable release returned in the inspected release list.
Correction guidance · Discussion is not enabled · Profile format 0.6.0-draft.1
Change history
2026-09-30 — Initial pinned-source component profile with setup route and integration boundaries.