Research map / Reinforcement learning / Imitation learning
Deep Imitative Models for Flexible Inference, Planning, and Control
Nicholas Rhinehart, Rowan McAllister, Sergey Levine
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 |
|---|---|---|---|
| Use an imitative trajectory-density prior with goal likelihood planning | CARLA dynamic driving · Town01 and Town02 | success rate | Improved |
| Use Gaussian Final-State Mixture goals instead of Final-State Indicator goals (ablation) | Imitative Model · CARLA Dynamic Navigation, Town02 | success rate: 84% -> 100% | Improved |
| Use Line Segment Final-State Indicator goals instead of Final-State Indicator goals (ablation) | Imitative Model · CARLA Dynamic Navigation, Town02 | success rate: 84% -> 88% | Improved |
| Use Region Final-State Indicator goals instead of Final-State Indicator goals (ablation) | Imitative Model · CARLA Dynamic Navigation, Town02 | success rate: 84% -> 88% | Improved |
| Add a smart waypointer that uses traffic-light state (ablation) | Imitative Model with Region Final-State Indicator · CARLA Dynamic Navigation, Town01 | red-light violations: 56% -> 2.8% | Improved |
| Add a smart waypointer that uses traffic-light state (ablation) | Imitative Model with Gaussian Final-State Mixture · CARLA Dynamic Navigation, Town01 | success rate: 92% -> 100% | Improved |
| Add test-time pothole costs (ablation) | Imitative Model with Gaussian Final-State Mixture · CARLA pothole avoidance, Town01 | potholes hit: 177/230 -> 10/230 | Improved |
| Add test-time pothole costs (ablation) | Imitative Model with Gaussian Final-State Mixture · CARLA pothole avoidance, Town02 | potholes hit: 82/154 -> 35/154 | Improved |
| Provide waypoints on the wrong side of the road (ablation) | Imitative Model with Gaussian Final-State Mixture · CARLA robustness, Town01 and Town02 | success rate: 100%; 70% | Mixed |
| Provide half of the waypoints as distractors (ablation) | Imitative Model with Gaussian Final-State Mixture · CARLA robustness, Town01 and Town02 | success rate: 70%; 50% | Mixed |
| Include traffic-light state in the trajectory model (ablation) | R2P2-based Imitative Model · CARLA driving | red-light stopping preference | Improved |
Closest papers by meaning
- Model Predictive Adversarial Imitation Learning for Planning from Observation (ICLR 2026 · similarity 0.90)
- Is Behavior Cloning All You Need? Understanding Horizon in Imitation Learning (NeurIPS 2024 · similarity 0.88)
- Zero-Shot Offline Imitation Learning via Optimal Transport (ICML 2025 · similarity 0.88)
- Interactive and Hybrid Imitation Learning: Provably Beating Behavior Cloning (NeurIPS 2025 · similarity 0.88)
- Imitation Learning from Observation with Automatic Discount Scheduling (ICLR 2024 · similarity 0.87)
- GO-DICE: Goal-Conditioned Option-Aware Offline Imitation Learning via Stationary Distribution Correction Estimation (AAAI 2024 · 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.
