Research map / Multi-agent systems and games / Human-AI and goal-conditioned
RoboTron-Sim: Improving Real-World Driving via Simulated Hard-Case
Baihui Xiao, Chengjian Feng, Zhijian Huang, Feng Yan, Yujie Zhong, Lin Ma
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 Scenario-aware Prompt Engineering (SPE) (ablation) | LLaVA-OneVision driving baseline · nuScenes planning ablation (no ego pose) | collision rate (%): 0.94 -> 0.79 | Improved |
| add Scenario-aware Prompt Engineering (SPE) (ablation) | LLaVA-OneVision driving baseline · nuScenes planning ablation (no ego pose) | boundary violation rate (%): 3.22 -> 2.68 | Improved |
| add Scenario-aware Prompt Engineering (SPE) (ablation) | LLaVA-OneVision driving baseline · nuScenes planning ablation (no ego pose) | L2 (m): 0.91 -> 0.86 | Improved |
| add Image-to-Ego (I2E) Encoder (ablation) | RoboTron-Sim with SPE · nuScenes planning ablation (no ego pose) | L2 (m): 0.86 -> 0.56 | Improved |
| add Image-to-Ego (I2E) Encoder (ablation) | RoboTron-Sim with SPE · nuScenes planning ablation (no ego pose) | collision rate (%): 0.79 -> 0.58 | Improved |
| add Image-to-Ego (I2E) Encoder (ablation) | RoboTron-Sim with SPE · nuScenes planning ablation (no ego pose) | boundary violation rate (%): 2.68 -> 3.02 | Hurt |
| add HASS simulated data to nuScenes | RoboTron-Sim · H2D scenarios (L2) | L2 (m) turn: 1.32 -> 0.64 | Improved |
| add HASS simulated data to nuScenes | RoboTron-Sim · H2D scenarios (night) | collision rate (%): 2.71 -> 1.56 | Improved |
| add HASS simulated data to nuScenes | VAD · E2D and H2D scenarios | L2 (m): 0.94 -> 0.93 | No clear effect |
| add HASS simulated data to nuScenes | MLLM baseline · H2D scenarios | L2 (m) rainy: 1.15 -> 0.90 | Improved |
| use hard-case HASS instead of distribution-matched GASS synthetic data (ablation) | RoboTron-Sim · H2D scenarios | L2 (m) night: 1.00 -> 0.81 | Improved |
| use hard-case HASS instead of distribution-matched GASS synthetic data (ablation) | RoboTron-Sim · H2D scenarios | L2 (m) turn: 1.21 -> 0.64 | Improved |
Closest papers by meaning
- ReSim: Reliable World Simulation for Autonomous Driving (NeurIPS 2025 · similarity 0.88)
- Hybrid-Driving: An Autonomous Driving Decision Framework Integrating Large Language Models, Knowledge Graphs and Driving Rules (AAAI 2025 · similarity 0.87)
- Generating Traffic Scenarios via In-Context Learning to Learn Better Motion Planner (AAAI 2025 · similarity 0.87)
- SafeAuto: Knowledge-Enhanced Safe Autonomous Driving with Multimodal Foundation Models (ICML 2025 · similarity 0.86)
- DiffScene: Diffusion-Based Safety-Critical Scenario Generation for Autonomous Vehicles (AAAI 2025 · similarity 0.86)
- NAVSIM: Data-Driven Non-Reactive Autonomous Vehicle Simulation and Benchmarking (NeurIPS 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.
