Embodied AI for the Physical World

We build agentic robots that perceive, reason, and act safely in unstructured environments.

Hierarchical, tool-using agents coordinating perception, planning, and control for long-horizon tasks.

Leverage multimodal foundation models for perception, understanding, and decision‑making.

Persistent visual, spatial, and semantic memory for robust grounding and planning.

Hybrid‑frequency control integrating global planning and reactive control.

High‑resolution tactile feedback enables delicate, reliable grasps.

We are hiring

Join us to push the frontier of embodied intelligence.