Research map / Reinforcement learning / Policy gradient and actor-critic
Continuous-Time Fitted Value Iteration for Robust Policies
Michael Lutter, Boris Belousov, Shie Mannor, Dieter Fox, Animesh Garg, Jan Peters
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 the adversary (robust rFVI instead of cFVI) | DP value iteration · Simulated Cartpole | reward: -024.2 -> -027.1 | Hurt |
| add the adversary (robust rFVI instead of cFVI) | DP value iteration · Physical Cartpole | reward: -143.7 -> -074.1 | Improved |
| add the adversary (robust rFVI instead of cFVI) | DP value iteration · Physical Cartpole | success rate: 73.3 -> 100.0 | Improved |
| add the adversary (robust rFVI instead of cFVI) | DP value iteration · Physical Furuta Pendulum | reward: -082.1 -> -278.0 | Hurt |
| add the adversary (robust rFVI instead of cFVI) | RTDP · Simulated Cartpole and Furuta pendulum | success | Hurt |
| use dynamic programming (uniform full-domain sampling) instead of RTDP | cFVI · Physical Furuta Pendulum | success rate: 00.0 -> 100.0 | Improved |
| increase n-step horizon of value target | cFVI/rFVI · simulated systems | convergence speed | Improved |
| use very long n-step horizons | cFVI/rFVI · Cartpole and Furuta pendulum | convergence | Hurt |
| increase adversary strength alpha | DP rFVI · simulated systems | reward | Hurt |
| increase adversary strength alpha | RTDP rFVI · simulated systems | task success | Hurt |
| use locally quadratic value network with sine/cosine transform instead of standard MLP | cFVI · Cartpole | learning curve | Improved |
| add domain randomization to PPO (uniform init) | PPO · Physical Furuta Pendulum | reward: -755.3 -> -080.6 | Improved |
Closest papers by meaning
- A Temporal Difference Method for Stochastic Continuous Dynamics (NeurIPS 2025 · similarity 0.86)
- Enhancing Value Function Estimation through First-Order State-Action Dynamics in Offline Reinforcement Learning (ICML 2024 · similarity 0.85)
- Time-Constrained Robust MDPs (NeurIPS 2024 · similarity 0.84)
- Robust Offline Reinforcement Learning with Linearly Structured $f$-Divergence Regularization (ICML 2025 · similarity 0.84)
- Efficient Value Iteration for s-rectangular Robust Markov Decision Processes (ICML 2024 · similarity 0.84)
- Continuous Q-Score Matching: Diffusion Guided Reinforcement Learning for Continuous-Time Control (NeurIPS 2025 · 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.
