Research map / Vision-language and multimodal / Robot action and navigation
E-VLA: Event-Augmented Vision-Language-Action Model for Dark and Blurred Scenes
Jiajun Zhai, Hao Shi, Shangwei Guo, Kailun Yang, Kaiwei Wang
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 event adapter | image-only VLA · Pick-Place 20 lux | success rate (%): 0 -> 90 | Improved |
| add event overlay on RGB | image-only VLA · Pick-Place 20 lux | success rate (%): 0 -> 60 | Improved |
| add event adapter instead of overlay | E-VLA · Pick-Place average | success rate (%): 80.8 -> 94.2 | Improved |
| add event adapter | image-only VLA · Pick-Place OOD low-light | success rate (%): 39.0 -> 75.0 | Improved |
| add event adapter | image-only VLA · Pick-Place 1000 ms blur | success rate (%): 0 -> 25 | Improved |
| add event overlay | image-only VLA · Sorting 1000 ms blur | success rate (%): 5 -> 32.5 | Improved |
| add low-light enhancement RetinexNet instead of events | image-only VLA · Pick-Place average | success rate (%): 47.5 -> 66.7 | Improved |
| add E2VID reconstruction | image-only VLA · Pick-Place average | success rate (%): 47.5 -> 35.8 | Hurt |
| use 2000-event window instead of 500 events (ablation) | event adapter · Pick-Place average | success rate (%): 67.5 -> 94.2 | Improved |
| use 2000-event window instead of 4000 events (ablation) | event adapter · Pick-Place average | success rate (%): 85.8 -> 94.2 | Improved |
| use event-count window instead of 20 ms time window (ablation) | event adapter · Pick-Place average | success rate (%): 82.5 -> 94.2 | Improved |
| add action-only stage before joint training (ablation) | event adapter · Pick-Place 30 lux | success rate (%): 50.0 -> 90.0 | Improved |
Closest papers by meaning
- ActionUNet: Improving Robustness of VLA Models with Efficient Multi-scale Fine-tuning (NeurIPS 2026 · similarity 0.90)
- ForceVLA: Enhancing VLA Models with a Force-aware MoE for Contact-rich Manipulation (NeurIPS 2025 · similarity 0.89)
- NS-VLA: Towards Neuro-Symbolic Vision-Language-Action Models (NeurIPS 2026 · similarity 0.89)
- VLA-REPLICA: A Low-Cost, Reproducible Benchmark for Real-World Evaluation of Vision-Language-Action Models (NeurIPS 2026 · similarity 0.89)
- VLA-Cache: Efficient Vision-Language-Action Manipulation via Adaptive Token Caching (NeurIPS 2025 · similarity 0.89)
- Event-Grounded Sparse Autoencoders for Vision-Language-Action Policies (NeurIPS 2026 · similarity 0.89)
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.
