A training-free inference layer that reweights stale memory and penalizes immediate reversals improves trajectory efficiency and keeps or slightly improves success on R2R and REVERIE with a GridMM-based navigator.
Target-driven structured transformer planner for vision- language navigation,
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.RO 1years
2026 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
DART-VLN: Test-Time Memory Decay and Anti-Loop Regularization for Discrete Vision-Language Navigation
A training-free inference layer that reweights stale memory and penalizes immediate reversals improves trajectory efficiency and keeps or slightly improves success on R2R and REVERIE with a GridMM-based navigator.