Grounding Robot Generalization in Training Data via Retrieval-Augmented VLMs
This work addresses the challenge of precise generalization evaluation in robotics, which is incremental as it builds on existing methods for policy analysis.
The paper tackles the problem of evaluating robot policy generalization by proposing RADAR, a framework that compares test tasks to training data using retrieval and vision-language models, demonstrating effectiveness in controlled experiments and scalability to large datasets with agreement to human benchmarks.
Recent work on robot manipulation has advanced policy generalization to novel scenarios. However, it is often difficult to characterize how different evaluation settings actually represent generalization from the training distribution of a given policy. To work towards more precise evaluation of generalization in robotics, we propose RADAR, a scalable framework for directly comparing test-time evaluation tasks to policy training data, to determine what form of policy generalization is required. RADAR consists of a two-stage pipeline: first, retrieval using generalist policy embeddings identifies which training examples are relevant for a given evaluation task. Next, vision-language models (VLMs) analyze the evaluation task against the retrieved data, outputting interpretable analysis on how they compare along a variety of axes, and an overall classification of what type of policy generalization is required. Through controlled experiments, we demonstrate that VLMs are effective at analyzing data for generalization, and that our retrieval step effectively identifies examples needed to make accurate classifications with respect to the training data. Furthermore, we scale RADAR to large-scale datasets, where we observe agreement with human-defined benchmark conditions from prior work. We provide demonstrations at radar-analysis.github.io.