Research map / Reinforcement learning / Policy gradient and actor-critic
A New, Physics-Informed Continuous-Time Reinforcement Learning Algorithm with Performance Guarantees
Brent A. Wallace, Jennie Si
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 RCI instead of cFVI | continuous-time RL control · Jet aircraft FPA loop | 1% settling time (s): 14.42 -> 7.17 | Improved |
| use RCI instead of cFVI at 25% modeling error | CT-RL · Jet aircraft FPA loop | 1% settling time (s): 15.20 -> 6.44 | Improved |
| use RCI instead of cFVI | CT-RL · Pendulum (nu=0%) | average return: -8.85 -> -12.17 | Hurt |
| use RCI instead of cFVI | CT-RL · DDMR (nu=0%) | average return: -4.12 -> -3.69 | Improved |
| use RCI instead of cFVI | CT-RL · Jet aircraft (nu=0%) | average return: -20.50 -> -8.26 | Improved |
| use RCI instead of Nominal LQ (learn from nonlinear dynamics and data) (ablation) | LQ design · Pendulum, Jet, DDMR with modeling error | policy optimality error: over 90% reduction in optimality error | Improved |
| use RCI instead of Nominal LQ at 25% modeling error (ablation) | LQ design · Pendulum | policy optimality error: 2.61 -> 0.13 | Improved |
| use RCI instead of Nominal LQ at 25% modeling error (ablation) | LQ design · Pendulum y=theta | 1% settling time (s): 2.12 -> 1.18 | Improved |
| use RCI instead of Nominal LQ at 25% modeling error (ablation) | LQ design · Jet aircraft FPA loop | overshoot (%): 1.11 -> 0.69 | Improved |
| use RCI instead of cFVI at 25% modeling error | CT-RL · Jet aircraft FPA loop | overshoot (%): 19.96 -> 0.69 | Improved |
| use RCI instead of cFVI | CT-RL · DDMR omega loop | 1% settling time (s): 12.81 -> 6.68 | Improved |
| use RCI instead of cFVI | CT-RL · DDMR | trajectory episodes required: RCI/FVI = 1/5,000,000 | Improved |
Closest papers by meaning
- Integral Performance Approximation for Continuous-Time Reinforcement Learning Control (ICLR 2025 · similarity 0.90)
- A Differential and Pointwise Control Approach to Reinforcement Learning (NeurIPS 2025 · similarity 0.88)
- Overcoming Slow Decision Frequencies in Continuous Control: Model-Based Sequence Reinforcement Learning for Model-Free Control (ICLR 2025 · similarity 0.88)
- Continuous Q-Score Matching: Diffusion Guided Reinforcement Learning for Continuous-Time Control (NeurIPS 2025 · similarity 0.86)
- Learning from Suboptimal Data in Continuous Control via Auto-Regressive Soft Q-Network (ICML 2025 · similarity 0.86)
- EvoControl: Multi-Frequency Bi-Level Control for High-Frequency Continuous Control (ICML 2025 · 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.
