Research map / Reinforcement learning / Visual and representation RL
Interferobot: aligning an optical interferometer by a reinforcement learning agent
Dmitry Sorokin, Alexander Ulanov, Ekaterina Sazhina, Alexander Lvovsky
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 radius randomization (ablation) | Interferobot DDQN agent · physical MZI | visibility: 0.74 ± 0.20 -> 0.96 ± 0.02 | Improved |
| add radius randomization (ablation) | Interferobot DDQN agent · physical MZI | return: 85 ± 69 -> 221 ± 54 | Improved |
| add white noise to pixels (ablation) | Interferobot DDQN agent · physical MZI | visibility: 0.82 ± 0.07 -> 0.96 ± 0.02 | Improved |
| add white noise to pixels (ablation) | Interferobot DDQN agent · physical MZI | return: 129 ± 43 -> 221 ± 54 | Improved |
| add brightness randomization (ablation) | Interferobot DDQN agent · physical MZI | visibility: 0.91 ± 0.04 -> 0.96 ± 0.02 | Improved |
| add brightness randomization (ablation) | Interferobot DDQN agent · physical MZI | return: 178 ± 39 -> 221 ± 54 | Improved |
| add duty cycle and trigger time randomization (ablation) | Interferobot DDQN agent · physical MZI | visibility: 0.89 ± 0.07 -> 0.96 ± 0.02 | Improved |
| add duty cycle and trigger time randomization (ablation) | Interferobot DDQN agent · physical MZI | return: 200 ± 42 -> 221 ± 54 | Improved |
| deploy on physical setup instead of simulator | Interferobot DDQN agent · simulation vs physical MZI | mean highest visibility: 0.986 -> 0.983 | Hurt |
| RL agent versus human expert using keyboard | Interferobot DDQN agent · physical MZI | mean highest visibility: 0.972 -> 0.983 | Improved |
| RL agent versus human manual knob alignment | Interferobot DDQN agent · physical MZI | time to fine-tuned alignment: 0.968 | Mixed |
Closest papers by meaning
- Kinetix: Investigating the Training of General Agents through Open-Ended Physics-Based Control Tasks (ICLR 2025 · similarity 0.83)
- RLIF: Interactive Imitation Learning as Reinforcement Learning (ICLR 2024 · similarity 0.83)
- MENTOR: Mixture-of-Experts Network with Task-Oriented Perturbation for Visual Reinforcement Learning (ICML 2025 · similarity 0.82)
- DMC-VB: A Benchmark for Representation Learning for Control with Visual Distractors (NeurIPS 2024 · similarity 0.81)
- Agentic Jigsaw Interaction Learning for Enhancing Visual Perception and Reasoning in Vision-Language Models (ICLR 2026 · similarity 0.81)
- Imitation Learning from Observation with Automatic Discount Scheduling (ICLR 2024 · similarity 0.81)
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.
