AI Paper Atlas

Research map / Multi-agent systems and games / Cooperative multi-agent RL

Beyond Rule-Based Agents: Active Markov Games for Realistic Multi-Agent Interaction in Autonomous Driving

Yuan Gui, Hongchen Luo, Jiao Wang, Liqi Qu

CVPR 2026 · Multi-agent systems and games › Cooperative multi-agent RL · approach: MARL benchmarks and grouping

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 changedWhereMeasuredEffect
train against a diverse opponent policy pool instead of rule-based opponentssingle-agent PPO baseline · CARLA unprotected intersectionscollision rate: 0.08 -> 0.60Hurt
add the opponent strategy pool (ablation)full AMG co-evolution method · CARLA (multi-agent)OOD success rate: 0.81 -> 0.98Improved
add the strategy inference module (ablation)full AMG co-evolution method · CARLA (multi-agent)OOD success rate: 0.84 -> 0.98Improved
add the map module (ablation)full AMG co-evolution method · CARLA (multi-agent)OOD success rate: 0.62 -> 0.98Improved
add the perfect perception module (ablation)full AMG co-evolution method · CARLA (multi-agent)OOD success rate: 0.54 -> 0.98Improved
add the strategy pool module (ablation)full AMG co-evolution method · CARLA (multi-agent)collision rate: 0.17 -> 0.02Improved
interactive co-evolution training instead of single-agent or rule-based trainingego agent · CARLA strategy diversitystrategy diversity: 0.3278 -> 0.5841Improved
increase TTC potential intensity beta from 0.5 to 0.9 (ablation)potential-field reward · CARLA hyperparameter sensitivityaverage return: 0.4: 118.3134, 0.6: 128.0711, 0.9: 106.6458Hurt
increase interaction weight to 0.7 (ablation)potential-field reward · CARLA hyperparameter sensitivityaverage return vs safety factor: return 135.3574 -> 128.1268; safety factor 3.78 -> 3.84Mixed
multi-agent IPPO baseline vs proposed method under diverse opponentsIPPO · CARLA unprotected intersections (diverse opponent)collision rate: 0.29 -> 0.02Improved

Closest papers by meaning

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.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy