Protocol Architecture

Sensora combines on-chain governance and incentives with off-chain data storage and processing to build a scalable protocol.

Layers Overview

  1. Blockchain Layer (Base)

    • Smart contracts for reputation, scoring, and incentives.

    • Transparent contributor registry.

    • Token and governance mechanisms.

  2. Data Layer

    • Datasets stored off-chain (IPFS, Arweave, Filecoin).

    • On-chain hash references ensure immutability.

    • Metadata mandatory (sensor type, timestamp, format, geo-tags).

  3. Scoring Layer (MCP-Sensora)

    • Evaluates each submission along Quality, Relevance, Diversity.

    • Uses peer-review + AI models for validation.

    • Weighted scoring impacts contributor reputation.

  4. Access Layer

    • APIs & SDKs for developers.

    • Dataset explorer for browsing/filtering.

    • Staking models for premium access.

    • 402-based micropayments for premium dataset unlocks

      • Free access to previews and low-resolution samples

      • Pay-per-use unlocks for full-resolution, annotations, or bulk data

      • Automated access for AI agents via HTTP 402 Payment Required flow

x402 ensures contributors are rewarded fairly while builders only pay for what they actually need.

MCP-Sensora Scoring Module

The MCP-Sensora module assigns a 0–100 score to each contribution:

  • Quality (40%)

    • Resolution, clarity, labeling accuracy, noise reduction.

    • Example: 4K drone video > blurry smartphone clip.

  • Relevance (40%)

    • Alignment with robotics use cases (navigation, mapping, perception).

    • Example: LiDAR scans of city streets > random vacation photos.

  • Diversity (20%)

    • Novelty of data (new locations, environments, sensor types).

    • Example: underwater robot sonar scans > another urban dataset.

Final score updates on-chain reputation and determines eligibility for rewards.

Contribution Flow

  1. Dataset submission

    • Contributor uploads dataset via Sensora UI/API.

    • Metadata is attached and hashed.

  2. Validation

    • MCP-Sensora evaluates dataset.

    • Reputation-weighted peers may upvote/downvote.

  3. On-chain registration

    • Hash + score stored on Base.

    • Reputation score updated.

  4. Incentives

    • Contributors receive reputation + token rewards.

    • Vesting prevents dumping.

PreviousMission & VisionsNextContributors & Builders

Last updated