Research on control of mobile robot cooperating with environment embedded sensors

在本研究中，我们的目标是完成智能空间与移动机器人的协作。这里我们把由分布式传感器和执行器组成的空间称为智能空间。智能空间的设计与实现构建了一个具有高度灵活性和可扩展性的系统我们利用RT-中间件将分布式设备构件化，构建了面向构件的智能空间。讨论了智能空间的观测功能和移动机器人导航功能的设计原则，并成功构建了一个易于添加、更改或删除传感器和配置功能的系统。利用来自智能空间的信息和机器人车载传感器进行移动机器人导航为了实现移动机器人在智能空间中的辅助，研究了移动机器人在生活环境中的导航方法。首先，我们提出了一种在智能空间中通过观察人类行走路径来获取环境主要路径的方法。其次，将主路径信息、自身位置信息与周围障碍物信息相结合，实现了移动机器人高效、自然的运动。智能空间和移动机器人观测信息的集成我们建议不仅使用分布式传感器，而且使用机器人车载传感器。提出了两种信息有效集成的方法，实现了移动机器人的导航。首先，我们证明了利用综合信息可以精确地跟踪运动目标和构建环境地图。然而，由于测量值之间的相关性，存在计算和连接负载增加的问题。为了解决这个问题，我们提出了融合方法。此外，我们还提出了一种利用粒子滤波获取机器人模型的方法，并将其集成到系统中，以达到提高鲁棒性和扩大适用范围的目的。

In this research, we aimed to accomplish the cooperation of intelligent space and mobile robots. Here we call the space composed of distributed sensors and actuators, intelligent space.1. Design and implementation of intelligent space to construct a system with high flexibility and extensibilityWe constructed component-oriented intelligent space by making distributed devices into component using RT-middleware. We discussed about design guidelines of observation function and mobile robot navigation function of the intelligent space and successed to build a system easy to add, change or delete the sensor and disposition functions.2. Navigation of mobile robot using information from intelligent space and robot onboard sensorsTo realize assistance of mobile robot in intelligent space, we investigated mobile robot navigation method in living environment. At first, we proposed a method to obtain main pathway of the environment by observing human walking path in intelligent space. Next, we showed that mobile robot can perform efficient and natural movement by integrating information of main pathway and own position with surrounding obstacle information.3. Integration of informations observed by intelligent space and mobile robotWe proposed to use not only distributed sensors but robot onboard sensors. We also proposed method to integrate both informations efficiently to navigate mobile robot. First, we showed that tracking moving objects and construction of environment map can be done precisely by using integrated information. However, there was a problem that the computation and connection load increases because of correlation between measurement values. To solve this problem, we proposed fusion methods. Moreover, we proposed a method to obtain robot model using particle filter and integrating it into the system to accomplish improvement of robustness and expansion of applicable scope.