Research map / Probabilistic and scientific modeling / Molecules and proteins
Testing the Safety of Self-driving Vehicles by Simulating Perception and Prediction
Kelvin Wong, Qiang Zhang, Ming Liang, Bin Yang, Renjie Liao, Abbas Sadat, Raquel Urtasun
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 learned perception/prediction noise (ContextNoise) to perfect-perception simulation | motion planning simulation · structured test track vehicle | AP @0.5 IoU: 77.2 -> 94.5 | Improved |
| add ContextNoise | simulation · structured test track pedestrian | AP @0.3 IoU: 47.4 -> 81.9 | Improved |
| add ContextNoise | simulation · structured test track bicyclist | AP @0.3 IoU: 47.8 -> 89.6 | Improved |
| add ContextNoise | PLT planner simulation · structured test track | collision simulation IoU: 70.3 -> 85.2 | Improved |
| add ContextNoise | ACC planner simulation · structured test track | l2 distance at 3.0s: 19.8 -> 15.1 | Improved |
| add ContextNoise | simulation · structured test track pedestrian | ADE at 60% recall: 36 -> 37 | No clear effect |
| add ContextNoise | simulation · structured test track bicyclist | ADE at 50% recall: 77 -> 79 | Hurt |
| increase ContextNoise training data from 5% to 100% of TOR4D train split (ablation) | ContextNoise · TOR4D val vehicle | FDE at 70% recall: 123 -> 70 | Improved |
| increase ContextNoise training data from 5% to 100% (ablation) | ContextNoise · TOR4D val vehicle | AP @0.5 IoU: 88.0 -> 91.7 | Improved |
| halve ContextNoise training data (100% to 50%) (ablation) | ContextNoise PLT · TOR4D val | collision simulation IoU: 74.9 -> 74.0 | Hurt |
| halve ContextNoise training data (100% to 50%) (ablation) | ContextNoise ACC · TOR4D val | collision simulation IoU: 64.3 -> 63.1 | Hurt |
| train ContextNoise on only 5% of data (ablation) | ContextNoise · TOR4D val bicyclist | ADE at 50% recall: 49 -> 126 | Hurt |
Closest papers by meaning
- Prioritizing Perception-Guided Self-Supervision: A New Paradigm for Causal Modeling in End-to-End Autonomous Driving (NeurIPS 2025 · similarity 0.86)
- What Probing Reveals about Autonomous Driving: Better Predictions Lead to Better Planning (NeurIPS 2026 · similarity 0.86)
- Monitoring of Perception Systems: Deterministic, Probabilistic, and Learning-Based Fault Detection and Identification (Abstract Reprint) (AAAI 2024 · similarity 0.86)
- NAVSIM: Data-Driven Non-Reactive Autonomous Vehicle Simulation and Benchmarking (NeurIPS 2024 · similarity 0.83)
- BAT: Behavior-Aware Human-Like Trajectory Prediction for Autonomous Driving (AAAI 2024 · similarity 0.83)
- DME-Driver: Integrating Human Decision Logic and 3D Scene Perception in Autonomous Driving (AAAI 2025 · similarity 0.83)
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.
