基于边界点过滤的多智能体快速协同探索算法  

Fast Multi-Agent Collaborative Exploration Algorithm Based on Boundary Point Filtering

作　　者：姚昌华 许浩 付澍 刘鑫 YAO Changhua;XU Hao;FU Shu;LIU Xin(School of Electronics and Information Engineering,Nanjing University of Information Science and Technology,Nanjing 210044,China;School of Microelectronics and Communication Engineering,Chongqing University,Chongqing 400044,China;College of information science and engineering,Guilin University of Technology,Guilin 541006,China)

机构地区：[1]南京信息工程大学电子与信息工程学院,南京210044 [2]重庆大学微电子与通信工程学院,重庆400044 [3]桂林理工大学信息科学与工程学院,桂林541006

出　　处：《信息网络安全》2023年第11期58-68,共11页Netinfo Security

基　　金：国家自然科学基金[61971439,61961010];国家自然科学联合基金[U22B2002];江苏省自然科学基金[BK20191329]。

摘　　要：针对多智能体在无先验知识的未知环境中进行自主协同探索任务优化问题,文章构建了多智能体协同探索未知环境的优化模型,并提出了一种基于障碍物边界点过滤(Multiple Agent Obstacle Frontier Point Filter,MAOFPF)的多智能体协同探索算法。该算法综合考虑边界点与障碍物相对分布情况,探索边界点过滤的距离阈值,进而优化多智能体探索任务选择和资源分配。仿真结果表明,在不同场景下,文章所提算法能够有效过滤边界点的干扰数据,保证系统的平稳运行,与原算法相比,系统抗扰动能力和泛化能力更高,地图覆盖率增长速度更快,平均效率提升了25.22%。In this research,the optimization problem of autonomous cooperative exploration tasks for multiple agents in an unknown environment without prior knowledge was addressed.To tackle this problem,an optimization model for multiple agents’cooperative exploration in an unknown environment was constructed,and a novel algorithm called multiple agent obstacle frontier point filter(MAOFPF)was proposed.The MAOFPF algorithm tooks into account the relative distribution between boundary points and obstacles,explored the distance threshold for filtering boundary points,and consequently improved the selection of exploration tasks and resource allocation for multiple agents.Simulation results demonstrate that the proposed algorithm effectively filters out interference data from boundary points in various scenarios,ensuring smooth system operation.As a result,the optimized system exhibits enhanced disturbance resistance and generalization ability.Furthermore,the algorithm achieves a higher map coverage rate compared to the original algorithm,with an average efficiency improvement of 25.22%.

关 键 词：自主探索 多智能体 MAOFPF算法 协同探索 

分 类 号：TP309[自动化与计算机技术—计算机系统结构]