Abstract: Inspired by the physics paradigm，discuss swarm-robots system capturing principle and method that is based on virtual force.Provide some performance metrics（such as stabilization time，layer number，distance between layers，robot number in a layer，distributing density and so on）；Analyze the relation between the capturing performance and the robots number，the virtual force.Put forward the layer force structure which not only quickens the capturing speed，but also improves the quality of capturing layer.Put forward the capturing method that robot moves around the target，the result of the experiment shows that this method can solve the local minima problem，and greatly improve the capturing speed and quality.Design a physics robot experiment to verify capturing method.At last，the directions of further research are proposed.

Key words: swarm-robots, artificial physics, capturing, virtual force

摘要： 启迪于物理学中的范例，探讨了基于虚拟力的群机器人围捕算法的原理和方法。制定了详细的系统性能评价指标（稳定时间、层数、层间距离、层中机器人个数、机器人密度等参数）。分析了围捕性能与机器人数量、虚拟力的关系；提出了目标对机器人作用力的分层结构思想，实验显示不仅加快了围捕速度，而且提高了围捕分层的质量；提出了机器人绕目标转动的围捕方法，解决了势场法中常见的局部极点问题，并且大大提高了围捕的速度和质量。设计了一个简单的实体机器人实验证实了围捕算法的可行性，指出了今后的研究方向。

关键词: 群机器人, 人工物理, 围捕, 虚拟力