Research map / Multi-agent systems and games / Cooperative multi-agent RL
Safe Continuous-time Multi-Agent Reinforcement Learning via Epigraph Form
Xuefeng Wang, Lei Zhang, Henglin Pu, Husheng Li, Ahmed H Qureshi
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 |
|---|---|---|---|
| remove the target loss from the critic objective (ablation) | EPI · Safe continuous-time MPE (Formation, Line, Target) | cumulative reward | Improved |
| add the value gradient iteration (VGI) loss to the critic (ablation) | EPI · Safe continuous-time MPE (Formation, Line, Target) | cumulative reward | Improved |
| add the HJB residual loss to the critic (ablation) | EPI · Safe continuous-time MPE (Formation, Line, Target) | cumulative reward | No clear effect |
| overweight the target loss in the critic objective (ablation) | EPI · MPE Line | average distance to target | Hurt |
| overweight residual or VGI loss (x20) (ablation) | EPI · MPE (Formation, Line, Target) | average distance to target | Hurt |
| use the epigraph reformulation instead of a collision penalty added to the cost (ablation) | EPI · Constrained coupled oscillator (value and action vs LQR ground truth) | value approximation accuracy | Improved |
| use a smaller auxiliary state z than z* during training (ablation) | EPI · MPE Formation and Line | violation rate vs cost | Hurt |
| use a larger auxiliary state z than z* during training (ablation) | EPI · MPE Formation and Line | violation rate vs cost | Mixed |
| use the computed optimal z* in training instead of randomly sampled z | EPI vs EPPO · Safe MPE and safe MA-MuJoCo | cost and constraint violation | Improved |
Closest papers by meaning
- Multi-Agent First Order Constrained Optimization in Policy Space (NeurIPS 2023 · similarity 0.90)
- Scalable Constrained Policy Optimization for Safe Multi-agent Reinforcement Learning (NeurIPS 2024 · similarity 0.89)
- Continuous-Time Value Iteration for Multi-Agent Reinforcement Learning (ICLR 2026 · similarity 0.88)
- Robust Multi-Agent Reinforcement Learning with Model Uncertainty (NeurIPS 2020 · similarity 0.88)
- HMARL-CBF – Hierarchical Multi-Agent Reinforcement Learning with Control Barrier Functions for Safety-Critical Autonomous Systems (NeurIPS 2025 · similarity 0.86)
- Sampling-Based Safe Reinforcement Learning (NeurIPS 2026 · 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.
