MARL 기반 교통 신호 제어 모델의 의사결정 과정을 다각도로 분석하여 이해를 높이고 효율적인 교통 관리 전략을 수립하는 것이 핵심 목표이다.
MARLens, a visual analytics system, provides comprehensive insights into the decision-making process and interactions of multi-agent reinforcement learning models applied to traffic signal control, enabling researchers to better understand and improve these models.
Introducing language grounding as an auxiliary learning objective enables multi-agent teams to learn human-interpretable communication protocols that maintain task performance and generalize to ad-hoc teamwork scenarios.
An artificial deep reinforcement learning agent can effectively nudge human-like conditional cooperator agents into higher levels of cooperation in public goods games by establishing cooperative social norms through its own contribution behavior.
Diversity confers resilience in natural systems, yet traditional multi-agent reinforcement learning techniques often enforce homogeneity. This work introduces a novel metric, System Neural Diversity (SND), to quantify behavioral heterogeneity in multi-agent systems, enabling the measurement and control of diversity.
본 연구는 무신호 교차로에서 다중 자율주행차량의 협력적 의사결정 문제를 해결하기 위해 주의 집중 메커니즘과 계층적 게임 선행 정보를 활용한 다중 에이전트 강화학습 알고리즘을 제안한다.
장기적 경험을 통해 에이전트들은 협력적인 이웃을 선별하고 선호하는 상호작용 전략을 개발하여, 네트워크 상호성을 높이고 전체적인 협력 수준을 향상시킨다.
LOQA는 상대방의 행동-가치 함수를 모델링하여 상호 협력을 유도하는 분산 강화학습 알고리즘이다.
Multi-agent synchronization tasks (MSTs) require precise timing and coordination among agents to achieve successful outcomes, posing significant challenges for existing multi-agent reinforcement learning (MARL) approaches.
Verco, a novel multi-agent reinforcement learning algorithm, enables agents to generate human-understandable verbal messages to enhance coordination and cooperation.