Advancing DialNav through Automatic Embodied Dialog Augmentation
For researchers in embodied AI and dialog navigation, this work provides a scalable data generation method and training techniques that significantly boost performance, though the gains are incremental as they build on existing DialNav and VLN frameworks.
The paper addresses the scarcity of training data for DialNav, an embodied dialog navigation task, by proposing an automatic data generation pipeline that creates the RAINbow dataset (238K episodes). Combined with dual-strategy training and a localization model, their method achieves a +89% improvement in success rate on Val Seen (58.24) and +100% on Val Unseen (29.05), establishing a new state of the art.
For embodied agents capable of physical interaction, the capability to create and understand dialog is crucial to ensure both safety and effectiveness. While DialNav~\cite{han2025dialnav} provides a framework for holistic evaluation of the dialog--execution loop in photorealistic indoor navigation, its performance remains limited by a critical scarcity of training data (2K episodes). To address this, we propose an automatic generation pipeline, and construct the \textbf{RAINbow} dataset, a large-scale training dataset with 238K episodes for DialNav. Our pipeline converts existing VLN datasets into multi-turn dialog and creates cost-efficient and high-quality dataset. Then, we introduce two additional complementary advances to unlock the data's full potential: (1) Dual-Strategy Training, a navigation training scheme to align the navigation training with the dynamic dialog-navigation loop, and (2) a localization model that leverages VLN knowledge. By combining these complementary solutions, our model substantially outperforms the baseline in success rate on both \textbf{Val Seen} (58.24, \textbf{+89\%}) and \textbf{Val Unseen} (29.05, \textbf{+100\%}) splits, establishing a new state of the art.