Research map / Continual learning / Online and replay-based
Lifelong Embodied Navigation Learning
Xudong Wang, Jiahua Dong, Baichen Liu, Qi Lyu, Lianqing Liu, Zhi Han
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add Knowledge Inheritance Strategy (KIS) (ablation) | Uni-Walker · LENL benchmark | SR (%): 60.3 -> 67.3 | Improved |
| add shared smoothing consolidation loss (SSC) (ablation) | Uni-Walker · LENL benchmark | SR (%): 59.7 -> 67.3 | Improved |
| add Experts Co-Activation Strategy (ECAS) (ablation) | Uni-Walker · LENL benchmark | SR (%): 58.1 -> 67.3 | Improved |
| add expert subspace orthogonality constraint (ESOC) (ablation) | Uni-Walker · LENL benchmark | SR (%): 63.5 -> 67.3 | Improved |
| add Navigation Specific Chain of Thought (NSCoT) (ablation) | Uni-Walker · LENL benchmark | SR (%): 51.1 -> 67.3 | Improved |
| use mixed matching instead of instruction-only matching in TAKA (ablation) | Uni-Walker · LENL benchmark | SR (%): 35.0 -> 67.3 | Improved |
| use mixed matching instead of observation-only matching in TAKA (ablation) | Uni-Walker · LENL benchmark | SR (%): 65.1 -> 67.3 | Improved |
| use mixed matching instead of observation-only matching in TAKA (ablation) | Uni-Walker · LENL benchmark | SPL (%): 62.7 -> 62.3 | Hurt |
| add KIS (ablation) | Uni-Walker · LENL benchmark | SR-F forgetting (%): 14.2 -> 4.3 | Improved |
| add ESOC (ablation) | Uni-Walker · LENL benchmark | SR-F forgetting (%): 9.8 -> 4.3 | Improved |
| add TAKA to a LoRA continual learner (Seq-FT 12 vs O-LoRA+TAKA 58) | LoRA continual learners · LENL benchmark | SR (%): 12 -> 58 | Improved |
| use Uni-Walker instead of previous best on unseen scenes | SD-LoRA+TAKA · LENL unseen scenes S16-S18 | SR (%): 57.0 -> 62.0 | Improved |
Closest papers by meaning
- Towards Learning a Generalist Model for Embodied Navigation (CVPR 2024 · similarity 0.89)
- TANGO: Training-free Embodied AI Agents for Open-world Tasks (CVPR 2025 · similarity 0.88)
- NavGPT: Explicit Reasoning in Vision-and-Language Navigation with Large Language Models (AAAI 2024 · similarity 0.88)
- NeSyC: A Neuro-symbolic Continual Learner For Complex Embodied Tasks In Open Domains (ICLR 2025 · similarity 0.88)
- NavBench: Probing Multimodal Large Language Models for Embodied Navigation (NeurIPS 2025 · similarity 0.87)
- Exploratory Retrieval-Augmented Planning For Continual Embodied Instruction Following (NeurIPS 2024 · similarity 0.87)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
