Research map / Reinforcement learning / Imitation learning
Learning by Watching
Check your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 |
|---|---|---|---|
| Learn from inferred states and actions of surrounding vehicles | Driving policy trained with one hour of data · NoCrash Regular, NoCrash Dense, and Original CARLA | success rate: 39%, 33%, and 32% | Helped |
| Add visibility map with late fusion (ablation) | LbW · NoCrash Regular, 1 hour | success rate: 64 ± 1 -> 92 ± 3 | Helped |
| Add visibility map with late fusion (ablation) | LbW · NoCrash Dense, 1 hour | success rate: 24 ± 1 -> 24 ± 0 | No clear effect |
| Add visibility map with late fusion (ablation) | LbW · Original CARLA, 1 hour | success rate: 74 ± 1 -> 92 ± 1 | Helped |
| Add visibility map with early fusion (ablation) | LbW · NoCrash Regular, 1 hour | success rate: 64 ± 1 -> 52 ± 1 | Hurt |
| Use late rather than early visibility fusion (ablation) | LbW + Visibility · NoCrash Regular, 1 hour | success rate: 52 ± 1 -> 92 ± 3 | Helped |
| Learn from inferred states and actions of surrounding vehicles | Driving policy trained with 10 minutes of data · NoCrash Regular | success rate: 46% with one hour of Ego data -> 52% with 10 minutes of LbW data | Helped |
| Add visibility map with late fusion (ablation) | LbW · Town 3 NoCrash Regular adaptation | success rate: 60 ± 1 -> 100 ± 0 | Helped |
| Add visibility map with late fusion (ablation) | LbW · Town 3 NoCrash Dense adaptation | success rate: 60 ± 0 -> 40 ± 0 | Hurt |
| Add waypoint refurbishment (ablation) | LbW + Visibility · Town 2 NoCrash Regular, NoCrash Dense, and Original CARLA | success rate: 64± 0 -> 80 ± 1; 32±3 -> 36± 0; 86± 1 -> 96±0 | Helped |
| Add waypoint refurbishment (ablation) | LbW + Visibility · Town 3 NoCrash Regular and NoCrash Dense | success rate: 40 ± 0 -> 60 ± 0; 30 ± 1 -> 40 ± 0 | Helped |
| Add waypoint refurbishment (ablation) | LbW + Visibility · Town 3 Original CARLA | success rate: 60 ± 0 -> 60 ± 0 | No clear effect |
Closest papers by meaning
- Learning From All Vehicles (CVPR 2022 · similarity 0.93)
- Prioritizing Perception-Guided Self-Supervision: A New Paradigm for Causal Modeling in End-to-End Autonomous Driving (NeurIPS 2025 · similarity 0.89)
- Knowledge Distillation from Single-Task Teachers to Multi-Task Student for End-to-End Autonomous Driving (AAAI 2024 · similarity 0.87)
- Uncertainty-Guided Never-Ending Learning to Drive (CVPR 2024 · similarity 0.87)
- Continuously Learning, Adapting, and Improving: A Dual-Process Approach to Autonomous Driving (NeurIPS 2024 · similarity 0.87)
- Transfer Your Perspective: Controllable 3D Generation from Any Viewpoint in a Driving Scene (CVPR 2025 · 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.