Scale Up Strategically: Learning Compositional Generalization via Bias-Aware Evaluation and Data Collection for Robotic Manipulation
- Published
- Source
- arXiv
- Paper number
- 709
- Field
- Robotics
- arXiv ID
- 2607.21582
Key points
- It builds a framework that decomposes language instructions into five factors, color, verb, object, size, and space, to measure bias.
- It finds a consistent bias structure across six base policies in the order color, object, spatial, verb, and size.
- It proposes two quantitative metrics, FDR for pairwise bias across factors and FDH for the full ranking.
- The bias diagnostics show that focusing data collection on weak factors improves generalization in real robots.
- It achieves better performance than the previous full dataset even when using only half of the demonstrations.
Paper links
External research summaries. These are not HDATF publications or measured product results.