SAIN: Structure-Aware Interactive Navigation with Active Dialogue Grounding for Mobile Robot
2026-08-10 • Robotics
Robotics
AI summaryⓘ
The authors address how robots navigate using unclear human instructions by allowing them to ask questions and remember the answers in a structured way. They created SAIN, a system that converts the robot's dialogue with humans into lasting memory about objects and places, rather than just temporary text clues. This helps the robot make better decisions about where to go and which objects to identify. Their approach improved navigation performance without needing special training for the task.
vision-language navigationinteractive dialogueambiguous instructionsembodied agentstructured memoryzero-shot learningnavigation policycorpus-level memoryfrontier rankingVL-LN IIGN benchmark
Authors
Yuhao Cao, Xiao Liu, Yang Xie, Lu Liu, Haoyao Chen
Abstract
Most existing vision-language navigation tasks assume that instructions are complete and unambiguous. However, real-world robots often encounter natural human instructions that are ambiguous, underspecified, or incomplete, requiring them to resolve such uncertainties through active questioning. Interactive Instance Goal Navigation (IIGN) requires an embodied agent to find the specific instance under an ambiguous category-level instruction through active dialogue. However, existing dialogue-enabled methods often consume oracle answers as transient textual context for immediate decisions, rather than persistent spatial or object-centric structured state. We present SAIN, a zero-shot framework that turns active dialogue into persistent navigation state. Instead of consuming oracle answers as one-step text hints, SAIN compiles them into target evidence, route-level corridor memory, and object-candidate labels. These states are stored in structured value, room, graph, and object memories, then consumed by a unified policy for frontier ranking and final target approach. On the VL-LN IIGN benchmark, SAIN improves SR from 20.2 to 25.4 and SPL from 13.07 to 14.17 over the strongest reported dialogue-enabled baseline, while requiring no task-specific policy training. The results support dialogue-to-state conversion as an effective zero-shot mechanism for long-horizon interactive instance navigation. Project website: https://zorattc.github.io/SAIN/