> For the complete documentation index, see [llms.txt](https://docs.idlenetwork.com/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.idlenetwork.com/idle-robotics-and-drones-marketplace/verification-and-proof-layer.md).

# Verification & Proof Layer

The hardest problem: How do we know a robot actually did the job?

Idle Network deploys the following technique:

1. **Sensor Proofs**: Robot signs sensor outputs (e.g., GPS logs, video hashes).
2. **Zero-Knowledge Proofs (ZKPs)**: Instead of uploading raw video (huge bandwidth/privacy risk), robot generates ZK proof that it processed footage correctly.
3. **Example**: “I detected 15 objects in this video stream” → proof verifies computation was correct without exposing video.
4. **Remote Auditing**: Third-party verifiers randomly challenge robots to perform micro-tasks (e.g., move camera, return sensor data).
5. **Hardware Attestation**: Use ARM TrustZone/TPM to ensure Idle Agent is running in secure enclave.

This verification layer is critical to prevent fake participation (e.g., a malicious actor pretending to run drones).