Robust Finetuning of Vision-Language-Action Robot Policies via Parameter Merging
This addresses the challenge of robustly adapting robot policies to new tasks without sacrificing generalist capabilities, though it is incremental as it builds on existing finetuning methods.
The paper tackles the problem of overfitting in finetuning generalist robot policies on new tasks, which causes loss of prior abilities and poor generalization. It shows that interpolating weights between finetuned and pretrained models preserves generalization, outperforming both models on out-of-distribution variations and enabling continual skill acquisition.
Generalist robot policies, trained on large and diverse datasets, have demonstrated the ability to generalize across a wide spectrum of behaviors, enabling a single policy to act in varied real-world environments. However, they still fall short on new tasks not covered in the training data. When finetuned on limited demonstrations of a new task, these policies often overfit to the specific demonstrations--not only losing their prior abilities to solve a wide variety of generalist tasks but also failing to generalize within the new task itself. In this work, we aim to develop a method that preserves the generalization capabilities of the generalist policy during finetuning, allowing a single policy to robustly incorporate a new skill into its repertoire. Our goal is a single policy that both learns to generalize to variations of the new task and retains the broad competencies gained from pretraining. We show that this can be achieved through a simple yet effective strategy: interpolating the weights of a finetuned model with that of the pretrained model. We show, across extensive simulated and real-world experiments, that such model merging produces a single model that inherits the generalist abilities of the base model and learns to solve the new task robustly, outperforming both the pretrained and finetuned model on out-of-distribution variations of the new task. Moreover, we show that model merging enables continual acquisition of new skills in a lifelong learning setting, without sacrificing previously learned generalist abilities.