Research map / Reinforcement learning / Visual and representation RL
CADRE: A Cascade Deep Reinforcement Learning Framework for Vision-Based Autonomous Urban Driving
Read the paperCheck 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 |
|---|---|---|---|
| use CADRE instead of LBC | Vision-based urban driving · NoCrash, Train town and Train weather, Dense traffic | success rate: 75 -> 82 | Helped |
| use CADRE instead of LBC | Vision-based urban driving · NoCrash, Test town and Test weather, Dense traffic | success rate: 12 -> 52 | Helped |
| use CADRE instead of IARL | Vision-based urban driving · NoCrash, Test town and Train weather, Dense traffic | success rate: 33 -> 61 | Helped |
| use CADRE instead of IARL | Vision-based urban driving · Vehicle avoidance | success rate: 69/81 -> 81/81 | Helped |
| use CADRE instead of LBC | Vision-based urban driving · Pedestrian avoidance | success rate: 73/78 -> 76/78 | Helped |
| add behavior-cloning control information to the perception module (ablation) | CoPM · Vehicle avoidance | success rate: 75/81 -> 81/81 | Helped |
| add behavior-cloning control information to the perception module (ablation) | CoPM · Pedestrian avoidance | success rate: 39/78 -> 76/78 | Helped |
| add co-attention between vision and behavior-cloning branches (ablation) | CoPM · Vehicle avoidance | success rate: 76/81 -> 81/81 | Helped |
| add co-attention between vision and behavior-cloning branches (ablation) | CoPM · Pedestrian avoidance | success rate: 68/78 -> 76/78 | Helped |
| add LSTM temporal modeling to basic CADRE (ablation) | CADRE PPO agent · Vehicle avoidance | success rate: 51/81 -> 71/81 | Helped |
| add LSTM temporal modeling to basic CADRE (ablation) | CADRE PPO agent · Pedestrian avoidance | success rate: 51/78 -> 61/78 | Helped |
| add distributed training with 4 workers to CADRE with LSTM (ablation) | CADRE PPO agent · Vehicle avoidance | success rate: 71/81 -> 81/81 | Helped |
Closest papers by meaning
- Prioritizing Perception-Guided Self-Supervision: A New Paradigm for Causal Modeling in End-to-End Autonomous Driving (NeurIPS 2025 · similarity 0.85)
- Graph-Based Prediction and Planning Policy Network (GP3Net) for Scalable Self-Driving in Dynamic Environments Using Deep Reinforcement Learning (AAAI 2024 · similarity 0.85)
- Drive-R1: Bridging Reasoning and Planning in VLMs for Autonomous Driving with Reinforcement Learning (AAAI 2026 · similarity 0.84)
- $AutoDrive\text{-}P^3$: Unified Chain of Perception–Prediction–Planning Thought via Reinforcement Fine-Tuning (ICLR 2026 · similarity 0.84)
- DriveAgent-R1: Advancing VLM-based Autonomous Driving with Active Perception and Hybrid Thinking (ICLR 2026 · similarity 0.84)
- Learning to Drive in New Cities Without Human Demonstrations (NeurIPS 2026 · similarity 0.84)
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.