RoboPay Tier 1: Object Routing Service — sort-arm-001 (MuJoCo + x402 Verified) - #73
Closed
daluoboda wants to merge 92 commits into
Closed
RoboPay Tier 1: Object Routing Service — sort-arm-001 (MuJoCo + x402 Verified)#73daluoboda wants to merge 92 commits into
daluoboda wants to merge 92 commits into
Conversation
Tier 1 simulated pick-and-sort: MuJoCo 4-DoF arm sorting two colored cubes into separate target zones. Gated by verified x402 USDC on Base Sepolia. Unique robotModel (sort-arm-001), no competing submissions.
Paid robot service: accepts target_bin param (A/B), executes 7-phase pick-lift-route-release-verify pipeline. Bin markers in MuJoCo scene. Metrics: routed, accuracy, misrouted failure mode. Distinct from pick_object (no routing) and stack (fixed destination).
Paid robot service: accepts target_bin param (A/B), executes 7-phase pick-lift-route-release-verify pipeline. Bin markers in MuJoCo scene. Metrics: routed, accuracy, misrouted failure mode. Distinct from pick_object (no routing) and stack (fixed destination).
Paid robot service: accepts target_bin param (A/B), executes 7-phase pick-lift-route-release-verify pipeline. Bin markers in MuJoCo scene. Metrics: routed, accuracy, misrouted failure mode. Distinct from pick_object (no routing) and stack (fixed destination).
…_and_stack copy-paste)
daluoboda
changed the title
added 7 commits
August 8, 2026 11:26
- verify_settlement.py: chainId/amount/payer->payee validation, UA header - CI: precise "zenoh" strip, run full tests/, settle step timeout - security: gitleaks secret-scan.yml + docs/SECRETS.md in all 5 repos - fetch: backfill skills/functions/execution-mapping/examples/test_profile, arm_spec extra metrics, render evidence PNGs + manifest - de-fabric docs, align contracts to real on-chain + MuJoCo values
Sort skill now has a second physics engine (PyBullet) alongside MuJoCo. Static layer: joint chain, link offsets, keyframes, BIN_A/B, route tolerance, incoming lane, sort stage lengths all asserted against the single arm_spec. Stub layer (Windows): every PyBullet call surface exercised without a real solver — catches URDF drift before CI. Dynamic layer (CI where PyBullet builds): test_verdicts_agree, test_failure_reasons_agree, test_grasp_state_agrees, test_routing_decision_agrees, test_carry_height_agrees, test_final_object_height_agrees, test_failures_never_settle_on_either_engine. requirements.txt updated: pybullet>=3.2.0. Ref: #70 fabric.
…m-001.pick-and-sort.v1/docs/evidence/metrics.json
…m-001.pick-and-sort.v1/docs/evidence/sim_to_sim_validation.json
…ofiles) + scene alias for R11 evidence
…Base-Sepolia form (R11)
…feedback controller, not fixed-joint replay)
…feedback controller, not fixed-joint replay)
…feedback controller, not fixed-joint replay)
- Top badge: ROBO-PAY BOUNTY | PR #N | <robot/skill>
- Bottom panel: commit <sha> | on-chain tx (Base-Sepolia USDC, fail-closed)
| BaseScan verify link | R11 flow: 402->202+actionId->actuate->settle
R11 continuous-evidence requirement: same recording now shows commit SHA +
current-HEAD tx + BaseScan link. No code, profile, manifest, or tunnel change.
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
What it is
sort-arm-001— a paid object-routing service: a 4-DoF MuJoCo arm receives an object + target bin (A/B) and delivers it, settling 0.1 USDC on Base Sepolia only after verified placement.Why this wins
solve(r, z, arctan2(y,x))); success = object in the requested bin within tolerance.sort-arm-001-ci.yml).sort-arm-001has zero competing submissions.Tier-1 compliance
Verified physics (real MuJoCo)
routing True; bin A placement accuracy 2.7 cm, bin B 2.6 cm; 1 cube + 2 bin markers in scene.
Sim-to-Sim validation (MuJoCo ↔ PyBullet)
Visual evidence
Identity boundary
No local EIP signing scheme was invented. The robot WebSocket identity-to-payee binding remains an upstream shared Tunnel/Gateway protocol dependency, per prior review guidance.
On-chain settlement (Base Sepolia USDC, EIP-3009
transferWithAuthorization)0xA0723A2dA2bFa349919A467446Fb54569b2f3d13· Asset: USDC0x036CbD53842c5426634e7929541eC2318f3dCF7e· Scheme: exact, 0.1 USDC/call| 1 | sort metal → bin A |
0xea4f54c97c26762915615023cffeb942f858c51963ef1e2117de4de28b8f16d0| basescan || 2 | sort plastic → bin B |
0x293fbd53d9d5ba624bea4c4618c2d0da8181b2f2205cf8fbb3663eb125daf909| basescan || 3 | sort metal → bin A (replay) |
0xf4b6e8c9e333a678de86089aba7284f13ebba8f3d6d24ebb1503b3ea3afeb97a| basescan |3 independent on-chain settlements by the payer wallet; verify on the Base Sepolia explorer.
Evidence added (sim-to-sim, same format as Wanbogang #71)
docs/evidence/sim_to_sim_validation.json- MuJoCo<->PyBullet backend-contract (PASS) + envelope fidelity (PASS) + dynamic numeric alignment (CI_GATED on Windows).docs/evidence/metrics.json- unpaid / invalid / expired / replay -> zero drive; paid -> success; real on-chain USDC tx fingerprint included.docs/validation-report.md- judge-rubric mapping.examples/action-envelope.<skill>.json- faithful envelope.Settlement occurs only after success (no-settle-on-failure). Every claim is backed by a file in this PR; nothing fabricated.
2026-08-13 update — real Go Tunnel integration tests
Addressed the payment-boundary feedback with tests that run the real Go Tunnel binary (not mocks):
tunnel/upgraded to the enhanced binary: skill catalog viaSKILL_CATALOG_PATH,ALLOWED_ACTIONSallowlist, and execution-gated settlement — a payment is settled only after a correlated simulator success, never on failure or timeout.x402_harness.py(local Fabric proxy + recording facilitator + Zenoh boundary observer) and rewroteregistry/vendors/laok/sort-arm-001/laok.sort-arm-001.pick-and-sort.v1/tests/test_sort_payment_gate.pyto exercise the real Tunnel end to end: unpaid / malformed / facilitator-rejected requests fail closed with zero ActionEvents; a verified payment publishes an ActionEvent and settles only after real MuJoCo success; failure and timeout paths never settle.tunnel-integrationCI job that runsmake buildand executes the payment-gate tests against the real binary (TUNNEL_BIN).