Do What? Teaching Vision-Language-Action Models to Reject the Impossible

TL;DR

This paper introduces IVA, a framework for vision-language-action models to detect, clarify, and respond to false-premise instructions in robotic tasks, improving robustness and understanding of user intent.

Contribution

The paper presents a novel unified framework, Instruct-Verify-and-Act (IVA), for detecting false-premise instructions and engaging in clarification, a significant advancement in VLA model robustness.

Findings

IVA improves false premise detection accuracy by 97.56%.

IVA increases successful responses in false-premise scenarios by 50.78%.

Constructed a large-scale dataset for training and evaluation.

Abstract

Recently, Vision-Language-Action (VLA) models have demonstrated strong performance on a range of robotic tasks. These models rely on multimodal inputs, with language instructions playing a crucial role -- not only in predicting actions, but also in robustly interpreting user intent, even when the requests are impossible to fulfill. In this work, we investigate how VLAs can recognize, interpret, and respond to false-premise instructions: natural language commands that reference objects or conditions absent from the environment. We propose Instruct-Verify-and-Act (IVA), a unified framework that (i) detects when an instruction cannot be executed due to a false premise, (ii) engages in language-based clarification or correction, and (iii) grounds plausible alternatives in perception and action. Towards this end, we construct a large-scale instruction tuning setup with structured language prompts and train a VLA model capable of handling both accurate and erroneous requests. Our approach leverages a contextually augmented, semi-synthetic dataset containing paired positive and false-premise instructions, enabling robust detection and natural language correction. Our experiments show that IVA improves false premise detection accuracy by 97.56% over baselines, while increasing successful responses in false-premise scenarios by 50.78%.