D-CLING attaches a zero-initialized trainable copy of a pre-trained navigation model backbone to learn in-domain depth cues without eroding prior generalization, enabling better real-world long-horizon navigation.
Tail: Task-specific adapters for imitation learning with large pretrained models
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.RO 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
D-CLING: Prior-Preserving Depth-Conditioned Fine-Tuning for Navigation Foundation Models
D-CLING attaches a zero-initialized trainable copy of a pre-trained navigation model backbone to learn in-domain depth cues without eroding prior generalization, enabling better real-world long-horizon navigation.