Probing a Vision-Language-Action Model for Symbolic States and Integration into a Cognitive Architecture

TL;DR

This paper investigates how a vision-language-action model encodes symbolic information and integrates it into a cognitive architecture to improve interpretability and robustness in robotic manipulation.

Contribution

It uncovers symbolic representations within OpenVLA's layers and demonstrates their integration into a cognitive architecture for enhanced interpretability.

Findings

High accuracy (> 0.90) in encoding object and action states across layers.

No observed pattern of earlier encoding of object states compared to action states.

Successful real-time state monitoring using symbolic representations.

Abstract

Vision-language-action (VLA) models hold promise as generalist robotics solutions by translating visual and linguistic inputs into robot actions, yet they lack reliability due to their black-box nature and sensitivity to environmental changes. In contrast, cognitive architectures (CA) excel in symbolic reasoning and state monitoring but are constrained by rigid predefined execution. This work bridges these approaches by probing OpenVLA's hidden layers to uncover symbolic representations of object properties, relations, and action states, enabling integration with a CA for enhanced interpretability and robustness. Through experiments on LIBERO-spatial pick-and-place tasks, we analyze the encoding of symbolic states across different layers of OpenVLA's Llama backbone. Our probing results show consistently high accuracies (> 0.90) for both object and action states across most layers, though contrary to our hypotheses, we did not observe the expected pattern of object states being encoded earlier than action states. We demonstrate an integrated DIARC-OpenVLA system that leverages these symbolic representations for real-time state monitoring, laying the foundation for more interpretable and reliable robotic manipulation.