Click to View CoCoMARL 2025 Website
Call For Papers
Stimulated by the increasing complexity of real-world problems, multi-agent reinforcement learning (MARL) has emerged as a promising approach for enabling autonomous agents to learn and adapt in dynamic environments. MARL broadly involves multiple autonomous agents learning to interact with each other in their environment, presenting a multitude of unique challenges and opportunities in both fundamental research and practical application.
In this workshop, we aim to focus particularly on MARL problems that require cooperation and/or coordination between multiple autonomous agents to be properly solved. This choice of specialization was made for multiple reasons. First and foremost, cooperative multi-agent systems are continuing to become more and more relevant - seeing effective use in robotic warehousing, space traffic management, and self-driving tasks - we believe this relevancy will only grow further with time. Second, MARL is a vast field of research with a multitude of interesting sub-areas and research problems, thus it would be most beneficial to the MARL community to dive deep into this one, particularly important category of MARL problems that has a wide range of open problems. Thus, the goal of this workshop is to bring together researchers and practitioners to discuss recent advances, challenges, and applications of MARL, with a particular focus on the following topics:
Multi-Agent Cooperation/Coordination: Methods and techniques for enabling agents to cooperate/coordinate and achieve goals in complex environments.
Inter Agent Communication: Approaches for facilitating communication between agents to enhance coordination and decision-making.
LLMs in MARL: Integration of Large Language Models (LLMs) into MARL systems to enable agents to understand and generate natural language, enhancing their ability to interact with humans and other agents.
Applications of MARL: Real-world applications of MARL in various domains, such as robotics, autonomous driving, finance, healthcare, and more.
Game Theory in MARL: Theoretical foundations and practical implications of applying game theoretic concepts in MARL settings.
Safety in MARL: Techniques and frameworks for ensuring the safety and reliability of MARL systems in critical environments.
Accepted Works
Click Here for a finalized list of papers accepted to the workshop.
Thank you to all authors, reviewers, and meta-reviewers who helped make this workshop possible.
Important Dates
Submission Deadline: 11:59 PM, May 17th, 2024 (AOE) May 10th, 2024 (AOE) April 26th, 2024
Acceptance Notification: May 31st, 2024 May 24th, 2024
Camera-ready Submission Deadline: July 26th, 2024 (AOE)
CoCoMARL 2024 Workshop: August 9th 2024, Amherst MA
Northeastern University
University of California Berkeley
LG AI Research
Massachusetts Institute of Technology
Northeastern University
Massachusetts Institute of Technology
University of Maryland, College Park
Massachusetts Institute of Technology
Northeastern University
The University of Manchester
Aditya Kapoor
Amin Me’marian
Anil B Murthy
Arrasy Rahman
Atrisha Sarkar
Bei Peng
Bilal Kartal
Daniel Melcer
Daphne Cornelisse
David Radke
Georgios Papoudakis
Ignacio Carlucho
Kyle Tilbury
Lei Yuan
Lucia Cipolina Kun
Manshi Limbu
Max Rudolph
Minsuk Chang
Pranay Pasula
Rupal Nigam
Sandy Tanwisuth
Satpreet Singh
Seth Karten
Sushmita Bhattacharya
Tianpei Yang
Victor Qin
Weixun Wang
Xian Yeow Lee
Xidong Feng
Xiyang Wu
Xueguang Lyu
Yang Li
Yuchen Xiao
Yuqian Fu
Zhehui Huang
Zun Li
Workshop Events and Features
☕ Topical Coffee Breaks - We will allocate multiple coffee break tables each marked with a sign denoting a particular topical focus relevant to the previous two talks. This gives participants an opportunity to discuss the previous talks, network with each other, and vote on questions to be asked during the panel discussion, as will be further explored in the next bullet point.
🗒️ "Wait Hold that Thought!" Sticky Notes - We will include ``Wait, Hold That Thought!" sticky notes as a method for attendees to write down their questions before they're forgotten. Attendees can then stick these questions to a wall section or poster board marked for each speaker or topic of focus. This open and physical system will allow attendees to see each other's questions, spark conversation between each other, and additionally vote on which questions they find most interesting. The most voted-for and or interesting questions will be addressed during the panel at the end of the day and discussed by both panelists and attendees.
🎤 Poster Elevator Pitches - We will give each presenter a brief opportunity (1-2 minutes depending on the number of posters) to pitch their work to everyone at the workshop. This will allow attendees to make more informed decisions about which posters to visit, give each accepted work a small yet well-deserved spotlight, and serve as a form of natural ice-breaker between presenting and non-presenting attendees.
🌈 Poster Passports - While the aforementioned elevator pitches will allow attendees to better determine which posters are most relevant or interesting to them, we still want to encourage broader and more interdisciplinary discussion. Therefore we will introduce a system of poster passports wherein each attendee will be rewarded for visiting posters from different areas of research being presented at the workshop.
🤔 Unsolved Problems in Collaborative MARL - We will kindly ask each speaker to dedicate some portion of their talk towards identifying some unsolved problem in collaborative MARL that they would like to see solved within the next 5-10 years. This will more effectively spark discussion during the topical coffee breaks panel discussion. We believe this will be a great opportunity for open discussion on the potential avenues of research as well as a great exercise to understand the different perspectives of the panelists on the topic.
❓Online Question and Polling Software - In the days leading up to the workshop, we will advertise a link on our workshop page to PollEverywhere, a software that allows users to anonymously ask questions and upvote other's questions. We will use this software to both drive engagement and enthusiasm for our workshop, and to help collect a set of questions for the speakers. One of the unique advantages of using PollEverywhere is that the collective question and voting results can be preserved for up to a year after their initial creation. This will allow attendees to revisit discussion points after the workshop for future use.
Call For Sponsors
We are actively looking for sponsors for the CoCoMARL workshop. Support us to gain valuable networking and exposure opportunities. See our prospectus document for sponsorship packages and benefits.