Robust Cooperative Algorithm Design for Robotic Sensor Networks and Experimental Verification

本计画的主要目的为：（1）基于视觉观测器的合作控制;（2）合作任务切换算法设计;（3）鲁棒合作控制算法设计;（4）建立一个机器人感测网路系统。关于主题（1）：我们成功地得出了关于实现成功合作的条件的理论结果。结果被接受为总理会议，在那里它收到了很高的评价，作为第一篇论文提供了一个非线性的基于非线性的协同控制方案的理论保证。这一结果也发表在国内期刊上。我们还将结果从针孔型摄像机模型扩展到全景摄像机模型，以克服视角问题，这是我们使用视觉传感器时的一个基本问题。研究结果还发表在一次国际会议和一份国内期刊上。关于主题（2）：我们已经成功地提出了一个基于所谓的共识协议的任务切换规律。这篇论文在一次重要的会议上获得了最佳演讲奖。此外，我们提出了另一种分布式任务切换算法的进化博弈论的基础上，这是发表在国际期刊上。关于主题（3）：我们想到了一个想法，即通过将博弈论与合作控制相结合来满足鲁棒合作。作为一个初步的，我们首先解决了问题（i）通过博弈论的方法，这是成功地接受了一个国际会议和国内期刊。此外，我们提出了一个学习算法，引导代理的最佳行动，即使没有任何先验知识的环境。一系列结果将在一次会议上发表，最近，期刊版本已被接受出版。关于主题（四）：我们成功地建立了一个机器人传感器网络系统，上面提到的大多数论文包括使用该系统的实验验证。

This project aimed at(1) Visual Observer-Based Cooperative Control,(2) Cooperative Task Switching Algorithm Design,(3) Robust Cooperative Control Algorithm Design,(4) Establishment of a Robotic Sensor Network System. On Topic(1): We succeeded in deriving theoretical results on conditions for achieving successful cooperation. The result was accepted for presentation at premier conferences, where it received a high evaluation as the first paper providing a nonlinear observer-based cooperative control scheme with theoretical guarantees. The result was also published in a domestic journal. We also extended the result from a pin-hole type camera model to a panoramic camera model so as to overcome the view angle issue which is an essential problem when we use a vision sensor. The result was also published in an international conference and a domestic journal. On Topic(2): We have succeeded in presenting a task switching law based on a so-called consensus protocol. The paper received a best presentation award at a premier conference. Moreover, we presented another distributed task switching algorithm based on evolutionary game theory, which was published in an international journal. On Topic(3): We thought of an idea that robust cooperation is met by integrating game theory with cooperative control. As a preliminary, we first addressed the problem in(i) via game theoretic approach, which was successfully accepted for an international conference and a domestic journal. Moreover, we presented a learning algorithm leading agents to the optimal actions even without any prior knowledge on environment. A series of the results will be presented at a conference and, recently, the journal version was accepted for publication. On Topic(iv): We succeeded in establishing a robotic sensor network system and most of the papers mentioned above include experimental verification using the system.