面向煤矿救援的群体机器人协作控制理论方法研究

Coal mine rescue robots are hugely demanded in the coal mine safety production and disaster rescue. Compared to more expensive and sophisticated individual robots, swarm robots are able to accomplish inherently distributed tasks in the presence of uncertainties. To this end, it is needed to develop self-organizing algorithms for swarm robotic systems, whose global behaviors emerged from local interactions of the robots are flexible and robust to the complex environment after a disaster that is unsafe for human to enter and try to find and rescue victims. To avoid the serious multi-path effect due to reflected waves of the mobile communication system signal in the roadways, more effective relative localization model and corresponding algorithm, which are very important to the system performance, are firstly developed. Then, the non-uniform rational B-spline model is used to represent the target distributed formation to be covered by the robots. At the same time, based on the coevolution and immune system optimization theory, a kind of niche particle swarm optimization model is developed to the multi-objective (victims) searching. Thereafter, inspired by the morphogenesis of biological organisms, a self-organizing distributed control model using a gene regulatory network (GRN) for swarm robotic systems is ultimately proposed to coordinately control individual robots to accomplish mine disaster environment detection and victims search and rescue. The study of this project will promote the theory development and practice application of swarm robots, and greatly improve the coal mine rescue efficiency and level of our country.

煤矿井下探测救援机器人在我国煤矿安全生产和救灾方面存在着巨大的需求。与更加昂贵和复杂的单体机器人相比，群体机器人能够本质上完成不确定性的分布式任务。针对煤矿事故后危险复杂的矿井环境，本项目研究巷道电磁波多径效应严重条件下基于移动无线传感器网络的群体机器人系统相对定位模型和算法，应用非均匀有理B样条模型构造群体机器人全区域覆盖的期望分布形态，基于协同进化与免疫系统优化原理构建群体机器人多目标搜索小生境微粒群模型，借鉴多细胞生物形态生成原理，建立基于基因调控网络的群体机器人分布式自组织控制系统模型，协调控制相对简单的个体机器人合作完成煤矿灾后环境探测、井下受困人员搜救等任务。本项目研究面向煤矿救援的群体机器人系统自组织分布式控制理论与方法，能为煤矿救援群体机器人的实际应用提供理论支持，有利于提高我国煤矿灾害的救援效率和水平。

煤矿井下探测救援机器人在我国煤矿安全生产和救灾方面存在着巨大的需求。与更加昂贵和复杂的单体机器人相比，群体机器人能够本质上完成不确定性的分布式任务。针对煤矿事故后危险复杂的矿井环境，本项目研究巷道电磁波多径效应严重条件下基于移动无线传感器网络的群体机器人系统相对定位模型和算法，应用非均匀有理B样条模型构造群体机器人全区域覆盖的期望分布形态，基于协同进化与免疫系统优化原理构建群体机器人多目标搜索小生境微粒群模型，借鉴多细胞生物形态生成原理，建立基于基因调控网络的群体机器人分布式自组织控制系统模型，协调控制相对简单的个体机器人合作完成煤矿灾后环境探测、井下受困人员搜救等任务。将差分GPS定位技术和RSSI无线定位技术相结合，提出了一种新的群机器人室外组合定位方法，消除了定位盲点，提高了定位精度；从物理学角度出发，抽象出了在复杂动态环境（包含静态和动态障碍物）下个体机器人基于简化虚拟受力的自主运动模型（SVF-model）；提出了一种基于简化虚拟受力模型的群机器人多目标搜索协调控制方法，相比于扩展粒子群算法(EPSO)，明显地减少了系统能耗和搜索耗时。本项目研究面向煤矿救援的群体机器人系统自组织分布式控制理论与方法，为煤矿救援群体机器人的实际应用提供理论支持，有利于提高我国煤矿灾害的救援效率和水平。