LIBERO-Pro extends the LIBERO benchmark suite with more complex manipulation tasks requiring multi-step reasoning, planning, and reusable skills. It is used to evaluate the generalization capability of embodied AI systems.
No structured details available.