Research map / Multi-agent systems and games / Human-AI and goal-conditioned
Generating High-Quality Explanations for Navigation in Partially-Revealed Environments
Gregory J. Stein
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 explanation-based training loss Lcomp (vs none) (ablation) | learned planner · Guided Maze | average distance: 19.17 -> 14.94 | Improved |
| add explanation-based training loss Lcomp (vs none) (ablation) | learned planner · University Buildings | average distance: 44.34 -> 42.46 | Improved |
| limit gradient to 4 most important subgoal properties instead of all (ablation) | learned planner · University Buildings | average distance: 43.52 -> 42.46 | Improved |
| limit gradient to 4 subgoal properties instead of all (ablation) | learned planner · Guided Maze | average distance: 14.94 -> 14.94 | No clear effect |
| train via explaining (4 properties) vs non-learned baseline | learned planner · University Buildings | average distance: 46.70 -> 42.46 | Improved |
| train via explaining (all properties) vs non-learned baseline | learned planner · Guided Maze | average distance: 19.45 -> 14.94 | Improved |
| learned baseline without Lcomp vs non-learned baseline | learned planner · Guided Maze | average distance: 19.45 -> 19.17 | No clear effect |
| intervene using explanation (4 properties) | learned planner · University Buildings worst 50 trials | average distance: 124.24 -> 93.77 | Improved |
| intervene using explanation (all properties) | learned planner · University Buildings worst 50 trials | average distance: 124.24 -> 92.81 | Improved |
| explain with all subgoal properties instead of 4 during intervention (ablation) | learned planner · University Buildings worst 50 trials | successful interventions: 37/50 -> 43/50 | Improved |
| intervened agent vs non-learned baseline | learned planner · University Buildings worst 50 trials | average distance: 85.05 -> 93.77 | Hurt |
Closest papers by meaning
- Counterfactual Vision-and-Language Navigation: Unravelling the Unseen (NeurIPS 2020 · similarity 0.88)
- Reasoning in Visual Navigation of End-to-end Trained Agents: A Dynamical Systems Approach (CVPR 2025 · similarity 0.88)
- NavGPT: Explicit Reasoning in Vision-and-Language Navigation with Large Language Models (AAAI 2024 · similarity 0.88)
- More Than Meets the Eye? Uncovering the Reasoning-Planning Disconnect in Training Vision-Language Driving Models (NeurIPS 2026 · similarity 0.87)
- Aux-Think: Exploring Reasoning Strategies for Data-Efficient Vision-Language Navigation (NeurIPS 2025 · similarity 0.86)
- Explaining Reinforcement Learning Agents through Counterfactual Action Outcomes (AAAI 2024 · 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.
