All projects
Safety & observability · Active
Physical AI Safety Observability
A runtime safety layer on Jetson AGX Thor that measures pipeline performance and inference latency, so the system can show when it is operating outside safe limits.
Hardware
The problem
A robot that works most of the time is not deployable. Operators need to know when perception or inference has degraded before it becomes a physical failure, and they need evidence they can review afterward.
Approach
- Instrument the full perception and inference pipeline on Jetson AGX Thor, not just the model.
- Track latency and throughput against explicit budgets.
- Use retrieval-grounded agents to help an operator interpret what the telemetry shows, with a human making the call.
Status
Active. Measurement is running on Thor.