Research map / Vision-language and multimodal / Robot action and navigation
Prioritized Semantic Learning for Zero-shot Instance Navigation
xinyu sun, Lizhao Liu, Hongyan Zhi, Ronghe Qiu, Junwei Liang
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 entropy-minimized goal view selection to ZSON (ablation) | ZSON · InstanceNav image-goal | SR: 12.7 -> 14.8 | Improved |
| add perspective reward relaxation to a model without the semantic perception module (ablation) | ZSON with goal view selection · InstanceNav image-goal | SR: 12.7 -> 10.8 | Hurt |
| add semantic perception module (SPM) to the agent (ablation) | ZSON · InstanceNav image-goal | SR: 12.7 -> 19.5 | Improved |
| add perspective reward relaxation on top of SPM and goal view selection (ablation) | PSL · InstanceNav image-goal | SR: 16.5 -> 22.0 | Improved |
| add perspective reward relaxation on top of SPM and goal view selection (ablation) | PSL · HM3D ObjectNav | SR: 35.0 -> 42.4 | Improved |
| add SPM and goal view selection together versus goal view selection alone (ablation) | PSL · InstanceNav image-goal | SR: 12.7 -> 16.5 | Improved |
| retrieve and weight image embeddings for text goal (semantic expansion inference) instead of direct text embedding (ablation) | PSL · InstanceNav text-goal | SR: 6.6 -> 16.5 | Improved |
| use ImageNav support set rather than limited-category IIN support set (ablation) | semantic expansion inference · InstanceNav text-goal | SR: 11.1 -> 12.4 | Improved |
| PSL versus LLM-based methods on ObjectNav, without LLM or mapping | PSL · HM3D ObjectNav | SR: 39.2 -> 42.4 | Improved |
| transfer OVRL ImageNav model to text goals via semantic expansion | OVRL · InstanceNav text-goal | SR: 3.7 | Hurt |
| PSL compared with a supervised model on image-goal SPL | PSL · InstanceNav image-goal | SPL: 11.8 -> 11.4 | Mixed |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- ZSON — ZSON: Zero-Shot Object-Goal Navigation using Multimodal Goal Embeddings (NeurIPS 2022) · better on 4 of 4 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- UniGoal: Towards Universal Zero-shot Goal-oriented Navigation (CVPR 2025 · similarity 0.89)
- BeliefMapNav: 3D Voxel-Based Belief Map for Zero-Shot Object Navigation (NeurIPS 2025 · similarity 0.89)
- RAO-Nav: Probing Omni-Language Models for Zero-shot Semantic Audio-Visual Navigation (NeurIPS 2026 · similarity 0.87)
- Instance-aware Exploration-Verification-Exploitation for Instance ImageGoal Navigation (CVPR 2024 · similarity 0.86)
- Imagine Before Go: Self-Supervised Generative Map for Object Goal Navigation (CVPR 2024 · similarity 0.86)
- Hierarchical Semantic-Augmented Navigation: Optimal Transport and Graph-Driven Reasoning for Vision-Language Navigation (NeurIPS 2025 · similarity 0.86)
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.
