Search papers, labs, and topics across Lattice.
This paper introduces DAGR, a novel approach that enhances goal-conditioned reinforcement learning by refining static embeddings into state-conditioned representations using multi-scale gated cross-attention. By incorporating a difference-aware mechanism, DAGR allows the model to better align goals with the current state, addressing the limitations of traditional state-independent encoders. Experimental results on OGBench demonstrate improved navigation performance, highlighting the importance of the gated residual in achieving these gains, while also noting that DAGR's effectiveness varies across different tasks.
DAGR transforms static goal representations into dynamic, state-aware embeddings, significantly boosting navigation performance in reinforcement learning tasks.
Goal-conditioned reinforcement learning hinges on how the goal is encoded. Contrastive, metric, temporal-distance, and information-theoretic encoders differ in objective. They still share one trait. None of them sees the current state. Such a state-independent embedding cannot mark which part of the goal still needs action. The policy must then recover that cue by inverting both encoders. We propose DAGR. It refines the static embedding of any late-fusion encoder into a state-conditioned one through multi-scale gated cross-attention. A near-identity gated residual preserves the base representation. Difference-aware Goal Cross-Attention then biases the attention scores using a per-token state-goal discrepancy map. On OGBench, DAGR improves navigation. Our ablations trace the gain to the gated residual, not to the difference bias that names the method. On manipulation and puzzle tasks it matches or falls below the base. DAGR is a structured refinement, not a universal improvement.
