What problem does it solve?
It helps you move from a broad robotics/embodied-AI direction to clear, benchmarkable research ideas with a concrete sim-first plan and explicit evaluation criteria, instead of producing generic ML concepts or untestable proposals.
Core Features & Use Cases
- Robotics-framed literature surveying: organizes prior work by embodiment, benchmark, task, observation/action interface, and sim2real constraints so gaps are actionable.
- Constrained idea generation and filtering: generates robotics-specific candidates and rejects weak ideas lacking benchmarks, baselines, measurable metrics, or credible sim-first pilots.
- Novelty checking and expert review packaging: prepares deep novelty verification and an external CoRL/RSS/ICRA-style review to validate contribution quality and missing evidence.
Use Case Example: If you say you want “sim2real ideas for bimanual manipulation,” it will first build a robotics landscape matrix, then propose top ideas with minimal sim-first validation packages, then move them through novelty checks and an external-style research review before producing a final report.
Quick Start
Use the skill with your robotics direction, for example: idea-discovery-robot "bimanual manipulation — sim-first, CoRL/RSS focus — no real-robot execution".
