Hi chvmp,
Posting this as a research-collaboration proposal. I'm Ido Yahalomi, maintainer of URML, an Apache 2.0 specification for substrate-neutral robot intent. Proposing a ChampAdapter that consumes CHAMP as a control-framework target (not a hardware target), routing URML programs onto any CHAMP-supported URDF (MIT Mini Cheetah, ANYmal, Spot, LittleDog, SpotMicroAI, OpenQuadruped).
URML's adapter pattern usually targets hardware (Spot via Boston Dynamics SDK, ANYmal via ANYbotics SDK). CHAMP is a different shape: a whole-body controller framework that abstracts the hardware geometry via URDF. A URML CHAMP adapter is the second path to the same hardware. URML's existing SpotAdapter (RFC-0043) and AnymalAdapter (RFC-0049) dispatch through proprietary SDKs; ChampAdapter dispatches through the open-source CHAMP controller on the same hardware. URML's user picks based on whether they need the proprietary SDK's full-feature surface or the open controller's customisability (gait research, sim-to-real).
This is proposal-only, posted as part of URML's Move #5 outreach (Tier B research-collab). No adapter code in this PR.
Two open questions stand out: (a) ROS 2 timeline (CHAMP is currently ROS 1 only — Kinetic / Melodic / Noetic), and (b) the per-customer-URDF manifest pattern (URML would ship example manifests covering CHAMP's supported URDFs; the deploying user adapts per their setup, mirroring URML's RFC-0070 HEBI per-customer-geometry approach).
Full RFC: https://github.com/URML-MARS/URML/blob/main/docs/rfcs/0077-mit-champ-outreach.md
Feedback we'd value
- ROS 2 timeline. Is CHAMP ROS 2 support planned?
- Per-platform vs platform-agnostic manifests. Per-URDF example manifests or a single parametric
champ_* manifest?
- Adapter home. URML repo, chvmp contributed example, both?
- SpotAdapter / ChampAdapter coexistence. Documented user guidance for picking between proprietary-SDK and CHAMP paths on the same hardware?
- ANYmal cross-coordination. Should URML coordinate with the open RFC-0049 (ANYmal) thread on CHAMP coexistence?
- Conformance lane. Open to a URML conformance line on
chvmp/champ README or in Discussions?
Thanks for CHAMP. The whole-body controller framework + URDF-parameterised approach is one of the cleanest abstraction shapes URML has encountered, and the MIT lineage gives it the institutional credibility URML's research-tier story needs.
Ido Yahalomi (URML maintainer, urml.dev)
Hi chvmp,
Posting this as a research-collaboration proposal. I'm Ido Yahalomi, maintainer of URML, an Apache 2.0 specification for substrate-neutral robot intent. Proposing a
ChampAdapterthat consumes CHAMP as a control-framework target (not a hardware target), routing URML programs onto any CHAMP-supported URDF (MIT Mini Cheetah, ANYmal, Spot, LittleDog, SpotMicroAI, OpenQuadruped).URML's adapter pattern usually targets hardware (Spot via Boston Dynamics SDK, ANYmal via ANYbotics SDK). CHAMP is a different shape: a whole-body controller framework that abstracts the hardware geometry via URDF. A URML CHAMP adapter is the second path to the same hardware. URML's existing
SpotAdapter(RFC-0043) andAnymalAdapter(RFC-0049) dispatch through proprietary SDKs;ChampAdapterdispatches through the open-source CHAMP controller on the same hardware. URML's user picks based on whether they need the proprietary SDK's full-feature surface or the open controller's customisability (gait research, sim-to-real).This is proposal-only, posted as part of URML's Move #5 outreach (Tier B research-collab). No adapter code in this PR.
Two open questions stand out: (a) ROS 2 timeline (CHAMP is currently ROS 1 only — Kinetic / Melodic / Noetic), and (b) the per-customer-URDF manifest pattern (URML would ship example manifests covering CHAMP's supported URDFs; the deploying user adapts per their setup, mirroring URML's RFC-0070 HEBI per-customer-geometry approach).
Full RFC: https://github.com/URML-MARS/URML/blob/main/docs/rfcs/0077-mit-champ-outreach.md
Feedback we'd value
champ_*manifest?chvmp/champREADME or in Discussions?Thanks for CHAMP. The whole-body controller framework + URDF-parameterised approach is one of the cleanest abstraction shapes URML has encountered, and the MIT lineage gives it the institutional credibility URML's research-tier story needs.
Ido Yahalomi (URML maintainer, urml.dev)