项目名称: 基于种群-虚拟物理力的多自主移动机器人主动嗅觉气味源定位策略研究
项目编号: No.61303183
项目类型: 青年科学基金项目
立项/批准年度: 2014
项目学科: 自动化技术、计算机技术
项目作者: 缪燕子
作者单位: 中国矿业大学
项目金额: 23万元
中文摘要: 通过使用移动机器人进行危险气味源定位问题已成为当前研究的热点。本项目以多机器人对危险气味源搜索定位应用为背景,以提高多气味源定位的成功率和速度为切入点,探索虚拟物理力和萤火虫群优化算法相结合的主动嗅觉算法。为了克服烟羽动态复杂特性的影响,同时避免陷入局部极值,拟提出基于虚拟物理力的主动嗅觉算法,在前期构造虚拟物理力的研究基础上,侧重研究采用启发式思想获得各分力的权重系数,以平衡三者分力之间的关系;利用萤火虫群优化算法可以同时计算多模函数的多优化问题的特性,拟提出基于种群-虚拟力的主动嗅觉算法,以进一步提高单个气味源定位的速度和鲁棒性,并解决多气味源定位的问题;基于启发式思想和气味质量通量计算方法,拟提出一种气味源确认算法,希望可以有效地排除发散的、不稳定局部最优点。最后利用FLUENT软件建立不同环境下的动态烟羽模型,用以验证提出的主动嗅觉算法的有效性和鲁棒性。
中文关键词: 主动嗅觉;虚拟力;改进烟花算法;点云数据;动态烟羽模型
英文摘要: The problem of dangerous gas source localization with mobile robot has become the focus of current research. Under the background of application on dangerous gas source localization with multi-robots, the project aims to study an active olfaction algorithm combined the virtual physics forces with the glowworm swarm optimization algorithm, being wished to improving the success rate and speed of gas source localization. In order to overcome the influence of plume dynamic feature and avoid the local extreme, based on the previous studies, the heuristic thought is introduced to get better weight coefficients of the component forces. As the glowworm swarm optimization algorithm could calculate multimode function of multi-optimization simultaneously, an active olfaction algorithm based on swarm and virtual physics would be put forward, which could improve the speed and robustness of single gas source localization effectively, and solves the problem of gas multi-source localization. In order to eliminate the divergent instability local peak, a gas localization algorithm would be proposed with heuristic thoughts and mass flux calculation method. Finally, the continuous dynamic plume model would be established with FLUENT software, where the effectiveness and robustness of proposed active olfaction algorithms would be ve
英文关键词: active olfaction;virtual physics force;OSL-FWA;point cloud data;turbulent plume model