1. Why we are interested in brain-inspired Physical AI?
Humans learn new skills from only a few examples, adapt instantly when something unexpected happens, and control a complex body with ease. Today’s robots, in contrast, often need huge amounts of data, struggle with situations they have not seen, and can fail in ways that are hard to predict.
The human brain solves these problems in its own way. It combines fast, reflexive control with slow, deliberate planning, predicts the consequences of its own actions, and keeps learning from every interaction with the world. We believe these principles can guide how we build Physical AI.
2. What are we working on?
We study how ideas from neuroscience and human motor control can be turned into architectures and learning methods for Physical AI, so that robots can:
- learn efficiently from limited data,
- adapt quickly to new objects, environments, and bodies, and
- act reliably in the physical world.
These methods form the foundation of our application-driven work, such as our Physical AI (VLA) projects.
More details will be shared as results are published.