Foundation models trained on cross-embodiment data (Open X-Embodiment, RT-X) now transfer semantic knowledge from vision-language models to low-level robot control, enabling zero-shot gener…

Pi runs its foundation model in the cloud, streaming actions to robots via API with real-time chunking that pipelines inference latency into the control loop, removing the need for expensiv…

Evaluation scales super-linearly with model capability; an AI agent that ingests multimodal failure data, diagnoses root cause across data/annotation/training, and proposes experiments woul…