Inductive Generalization for Robotic Manipulation
For robotic manipulation researchers, this work identifies a critical failure mode in current visuomotor policies that must be solved to achieve general-purpose robots.
The paper introduces 'inductive generalization' as a new evaluation paradigm for robotic manipulation policies, testing out-of-distribution task variants rather than interpolation. It shows that state-of-the-art Vision-Language-Action models fail these tests, exposing a fundamental limitation not addressed by scaling data or models.
Understanding the generalization capabilities of visuomotor policies is essential in the development of capable robotic agents. Generalizable models learn structures that transfer across domains. However, in practice, visuomotor policies test performance by interpolation on known distributions using unstructured domain shifts (e.g. lighting, clutter, diverse objects). We argue that to measure generalization capabilities we must instead test the inductive capacity of policies on progressively harder, out-of-distribution task variants. We call this inductive generalization, drawing directly on how axis-based evaluation has revealed inherent generalization limitations in language models (e.g. sequence length, counting) arXiv:2502.00197 . We provide a reusable and formal evaluation protocol for measuring inductive generalization in any manipulation policy, and establish baselines showing that existing paradigms fail this test; e.g. SoTA Vision-Language-Action models and find that policies that appear to generalize to prior domain shifts (distractors, etc) fail inductive generalization tests. These results expose a class of learning challenges orthogonal to those addressed by data and model scaling in robot learning, yet are imperative to solve in order to realize general purpose robots.