Search papers, labs, and topics across Lattice.
This paper introduces a novel approach to Vision-Language Navigation (VLN) under bandwidth constraints, focusing on a method called hindsight gating that identifies critical communication moments after navigation failures. The key finding reveals that agents communicate more effectively when they are confident and early in the episode, leading to improved trajectory alignment and navigation efficiency. By leveraging recurrent hidden-state alignment, the proposed method achieves significant gains in alignment efficiency compared to traditional gating methods, suggesting a paradigm shift in communication strategies for bandwidth-limited agents.
Agents communicate more effectively when they are confident and early in the episode, challenging conventional wisdom about uncertainty-driven communication.
Most cooperative Vision-Language Navigation (VLN) methods assume unlimited communication, not considering real-world applications where bandwidth is restricted and information efficiency is critical. We introduce \textbf{bandwidth-constrained cooperative VLN} and propose \textbf{hindsight gating}: a lightweight supervised gate that labels communication-critical steps post-hoc from navigation failures, avoiding the high variance of REINFORCE. Contrary to the intuition that agents should communicate when uncertain, we observe a consistent counter-intuitive pattern: trained gates fire predominantly in early episode steps and more often when agents are confident, across all budget levels ($B \in \{1,3,5\}$). We explain this through \textbf{recurrent hidden-state alignment}: early communication injects grounded trajectory representations that persist and compound through subsequent Gated Recurrent Unit (GRU) updates, achieving $+0.072$ cumulative alignment gain with $B{=}3$ transmissions, approaching unconstrained communication ($+0.078$) at 260\% greater alignment efficiency than random gating ($+0.020$) and 320\% greater efficiency than entropy-based gating ($+0.017$). Our results establish a new communication regime for bandwidth-limited embodied agents: synchronise representations early, navigate independently later. Our codebase is available at: https://github.com/AravG13/bandwidth-constrained-cooperative-vln
