Arrow Research search

Author name cluster

Zijian Ma

Possible papers associated with this exact author name in Arrow. This page groups case-insensitive exact name matches and is not a full identity disambiguation profile.

2 papers
1 author row

Possible papers

2

AAMAS Conference 2026 Conference Paper

GCMRBench: Goal-Conditioned Multi-Robot Environments and Benchmarks for Advancing Offline Multi-Agent Reinforcement Learning

  • Chenxing Li
  • Chin-jui Chang
  • Yinlong Liu
  • Zijian Ma
  • Jan Seyler
  • Shahram Eivazi

Research in multi-agent reinforcement learning (MARL) has focused on developing algorithms to address challenges posed by agents’ diverse goals, collaboration, and competition in complex environments. Extending these algorithms to Offline MARL (OMARL), together with the utilization of large-scale offline datasets, has increasingly been recognized as a promising approach toward safe, efficient, and rapid deployment in real-world scenarios. However, most existing studies train and evaluate OMARL in environments primarily designed for game-based scenarios. As a result, the potential of OMARL in domains such as robotics remains an open question. To bridge this gap, we introduce GCMRBench, a goalconditioned multi-agent simulation environment tailored for dualarm robotic tasks and the evaluation of offline multi-agent algorithms, thereby facilitating their deployment in practical robotic applications.

PDF

AAAI Conference 2025 Conference Paper

WebPilot: A Versatile and Autonomous Multi-Agent System for Web Task Execution with Strategic Exploration

  • Yao Zhang
  • Zijian Ma
  • Yunpu Ma
  • Zhen Han
  • Yu Wu
  • Volker Tresp

LLM-based autonomous agents often fail to execute complex web tasks that require dynamic interaction, largely due to the inherent uncertainty and complexity of these environments. Existing LLM-based web agents typically rely on rigid, expert-designed policies specific to certain states and actions, lacking the flexibility and generalizability needed to adapt to unseen tasks. In contrast, humans excel by exploring unknowns, continuously adapting strategies based on new observations, and resolving ambiguities through exploration. To emulate human-like adaptability, web agents need strategic exploration and complex decision-making. Monte Carlo Tree Search (MCTS) is well-suited for this, but classical MCTS struggles with vast action spaces, unpredictable state transitions, and incomplete information in web tasks. In light of this, we develop WebPilot, a multi-agent system with a dual optimization strategy that improves MCTS to better handle complex web environments. Specifically, the Global Optimization phase involves generating a high-level plan by breaking down tasks into manageable subtasks, continuously refining this plan through reflective analysis of new observations and previous subtask attempts, thereby focusing the search process and mitigating challenges posed by vast action spaces in classical MCTS. Subsequently, the Local Optimization phase executes each subtask using a tailored MCTS designed for complex environments, effectively addressing uncertainties and managing incomplete information by iteratively refining decisions based on new observations. Experimental results on WebArena and MiniWoB++ demonstrate the effectiveness of WebPilot. Notably, on WebArena, WebPilot achieves SOTA performance with GPT-4, achieving a 93% relative increase in success rate over the concurrent tree search-based method. WebPilot advances autonomous agents, enabling more reliable decision-making in practical environments.

PDF Details DOI
v2026.09.13