Humanoid Labs

Inspiration

Autonomous robotics is advancing rapidly, but ensuring reliability and accountability is becoming a major challenge. As autonomous systems take on more critical roles, we need transparent ways to verify what they were running and why they behaved as they did.

What it does

EigenCompute enables verifiable robotics by saving sensor data, AI model hashes, and outputs onto the blockchain. This allows post-verification of exactly what was executed on an autonomous robot—like a digital black box for AI, but cryptographically verified and tamper-proof.

How we built it

A ROS node records event timestamps, model hashes, TEE attestation receipts, and Merkle roots of logged data. Raw data is stored on IPFS or secure storage, while Merkle roots, attestations, and optional zk-proof receipts are anchored onchain via an EigenCompute AVS. A simple web UI allows fleet operators and auditors to view and verify these logs in real time.

Challenges we ran into

Finding the most effective way to communicate and visualize the concept of verifiable autonomy.

What's next for Humanoid Labs

Integrate EigenCompute into our own robots and open-source a standardized framework for verifiable autonomy that any robotics platform can adopt.

Built With

• eigenvalue
• rust