Proof-of-Physical-Work (PoPW)
Evidence Primitives
Evidence is constructed as a Merkle tree over signed leaves:
GPS trail:
(t, lat, lon, alt, DOP, speed)with device signature and anti-replay nonceIMU/LiDAR: Downsampled statistics and occasional raw frames for audits
Media hashes: Photo/video chunk hashes with capture-time and beacon challenge codes
Attestation quotes: TEE/TPM quotes binding firmware and session key [5]
Beacon responses: TLS/BLE challenge-response from environment beacons where available
Artifacts are referenced by a CID index included as leaves. The robot produces a public manifest:
Signed by the session key and optionally by the device key.
Deterministic Scoring
Let q ∈ [0,1] be the quality score computed by a public function Score(M, artifacts) defined in the job criteria.
Example (area-coverage jobs):
With Σ λi = 1 and each term in [0,1].
Challenge Protocol
A challenger posts bond Bch and claims a defect with a machine-checkable predicate (e.g., missing tile proof, inconsistent timing). The contract opens a short evidence window for oracle re-checks.
Outcomes:
Valid challenge: Refund
Bchand pay rewardρBchfrom slashes; revert payment or claw back (if not yet released).Invalid challenge: Burn
βBchand pay(1−β)Bchto the robot and treasury.
The game is tuned so truthful challenges have positive expected value when defect probability exceeds a threshold, discouraging spam.
Payoff Analysis with Numerics
Parameters:
Bond
Bch = 100 MORPHReward share
ρ = 0.4Burn fraction on failed challenge
β = 0.3Robot stake
s = 500 MORPHOracle aggregate stake
so = 10,000 MORPHSlashes:
σr = 0.20,σo = 0.05when challenge succeeds
Table 2: Challenge payoff matrix (example numbers)
Outcome | Challenger | Robot | Oracles |
|---|---|---|---|
Valid challenge |
|
|
|
Invalid challenge |
|
|
|
Dispute flow:
After quorum attestation, a challenge window
Tcopens.Valid challenge → transition to Challenge Valid → then Resolve with pay/slash.
Timeout or invalid challenge → No Challenge / Invalid → Resolve.
All payouts and slashes follow rules in Section 4.7 and Settlement invariants.
Threats and Mitigations
Table 3: Threats and mitigations
Threat Vector | Mitigation |
|---|---|
Sensor spoofing (synthetic or replayed packets) | TEE-bound session keys, nonces, timing analysis, cross-sensor checks |
Locality forgery (GPS-only tricks) | Multi-sensor fusion, WiFi/BLE fingerprints, visual place recognition |
Oracle collusion (robot + oracle bribe) | VRF sampling, stake caps, audits, slashable oracle stake |
Late reveal (reveal after observing oracles) | Commit-then-reveal acceptance with timeout |
Privacy leakage (over-sharing raw media) | Commitments on-chain, encrypted artifacts, selective access |
Censorship (oracle exclusion of a robot) | Multiple oracles, permissionless relays, L2 failover |