This paper considers the problem of decentralized goal assignment and trajectory generation for multi-robot networks when only local communication is available, and proposes an approach based on methods related to switched systems and set invariance. A family of Lyapunov-like functions is employed to encode the (local) decision making among candidate goal assignments, under which a group of connected agents chooses the assignment that results in the shortest total distance to the goals. An additional family of Lyapunov-like barrier functions is activated in the case when the optimal assignment may lead to colliding trajectories, maintaining thus system safety while preserving the convergence guarantees. The proposed switching strategies give rise to feedback control policies that are computationally efficient and scalable with the number of agents, and therefore suitable for applications including first-response deployment of robotic networks under limited information sharing. The efficacy of the proposed method is demonstrated via simulation results and experiments with six ground robots.
Decentralized Heterogeneous Multi-Robot Collaborative Exploration for Indoor and Outdoor 3D Environments
College of Artificial Intelligence, Harbin Institute of Technology (Shenzhen)(哈尔滨工业大学(深圳)人工智能学院);Department of Mechanical and Automation Engineering, The Chinese University of Hong Kong(香港中文大学机械与自动化工程系)
Learning Multi-Agent Task Assignment and Navigation in the Factory: from Simulation to Real Robots
INIT Robots Lab, École de technologie supérieure(INIT机器人实验室,高等工程技术学院);MISTLab, Department of Computer and Software Engineering, Polytechnique Montréal(MIST实验室,蒙特利尔理工学院计算机与软件工程系)
NeuroMesh: A Unified Neural Inference Framework for Decentralized Multi-Robot Collaboration
New York University(纽约大学);U.S. Army Combat Capabilities Development Command, Army Research Laboratory(美国陆军作战能力发展司令部,陆军研究实验室);Vanderbilt University(范德比尔特大学);University of California Berkeley, Department of Electrical Engineering and Computer Sciences(加州大学伯克利分校,电气工程与计算机科学系)
CoMMa: Contribution-Aware Medical Multi-Agents for Decentralized Oncology Decision Support
Center for Advanced Medical Computing(先进医学计算中心);Department of Radiology, Massachusetts General Hospital(放射科,马萨诸塞总医院);Harvard Medical School(哈佛医学院);Department of Radiation Oncology, Yonsei University College of Medicine(放射肿瘤科,延世大学医学院);Yonsei University(延世大学);Institute for Innovation in Digital Healthcare(数字医疗创新研究所);Interventional Radiology Academic Medical Centers, Mass General Brigham(介入放射学学术医疗中心,马萨诸塞总医院 Brigham)
Multi-agent In-context Coordination via Decentralized Memory Retrieval
National Key Laboratory of Novel Software Technology, Nanjing University(南京大学计算机软件新技术国家重点实验室);School of Artificial Intelligence, Nanjing University(南京大学人工智能学院);Tencent(腾讯)
Shapley Counterfactual Credits for Multi-Agent Reinforcement Learning
Zhejiang University(浙江大学);The Chinese University of Hong Kong, Shenzhen(香港中文大学(深圳));Shenzhen Institute of Artificial Intelligence and Robotics for Society(深圳市人工智能与机器人研究院)
DMAPF: A Decentralized and Distributed Solver for Multi-Agent Path Finding Problem with Obstacles
多智能体路径寻找(MAPF)是一个问题,找到一个移动的序列,以达到他们的分配位置没有冲突的代理。集中式算法通常会给出最优解,但在不采用各种技术的情况下很难扩展-通常会牺牲最优性;但解决代理数量超过一千的MAPF问题仍然是一个挑战。为了解决可扩展性问题,我们提出了DMAPF -一个分散和分布式的MAPF求解器,这是我们最近发表的工作,ros-dmapf的延续。我们解决了ros-dmapf的问题,它(i)只适用于没有障碍的地图;和(ii)有一个低的成功率与密集的地图。给定一个MAPF问题,ros-dmapf和DMAPF都在空间上将地图划分为子问题,但后者进一步将每个子问题划分为称为区域的不连通区域。每个子问题被分配给一个分布式求解器,然后单独创建一个抽象的计划-一个代理需要访问的区域序列-为每个代理,并交错代理迁移与移动规划。Answer Set Programming以其在小而复杂的问题中的性能而闻名,用于许多部分,包括问题划分,抽象规划,迁移的边界分配和移动规划。机器人操作系统用于促进求解器之间的通信,并使其有机会与机器人系统集成。DMAPF在求解器之间引入了一种新的交互协议,以及一些机制,这些机制共同导致了更高的成功率和更好的解决方案质量,而不会牺牲太多的性能。我们实现了DMAPF,并通过与其他国家的最先进的MAPF求解器进行比较,实验验证,结果表明,我们的系统实现了更好的可扩展…
Secure Distributed/Federated Learning: Prediction-Privacy Trade-Off for Multi-Agent System
University of Southern California(南加州大学)
分散学习是一种有效的新兴范式,用于提高多个有界计算代理的计算能力。在大数据时代,在分布式和联邦学习(DL和FL)框架内执行推理,中央服务器需要处理大量数据,同时依赖各种代理执行多个分布式训练任务。考虑到分布式计算的拓扑结构,隐私已经成为一个首要的问题。此外,假设代理的信息处理能力有限,需要一个复杂的\textit{privacy-preserving decentralization}来确保有效的计算。为此,我们研究的\textit{privacy-aware server to multi-agent assignment}问题受到信息处理的限制与每个代理,同时保持隐私和确保学习信息的消息,通过分布式私人联邦学习(DPFL)的方法收到代理全球终端。为了找到一个分散的计划,两个代理系统,我们制定了一个优化问题,平衡隐私和准确性,考虑到与每个代理相关联的压缩约束的质量。我们提出了一个迭代收敛算法交替自洽方程。我们还对所提出的解决方案进行了数值评估,以显示隐私预测的权衡,并证明了新方法在确保DL和FL隐私方面的有效性。