【作者】 王硕；

【导师】 张云洲；

【作者基本信息】 东北大学 ， 模式识别与智能系统， 2014， 硕士

【摘要】 救援机器人能够代替人员深入火场、矿井等危险的现场进行救援,受到了国内外研究机构的广泛关注。然而,传统救援机器人受到遥控距离短、感知范围有限、定位精度不准确等限制,对其实际应用造成了很大的障碍。本文将移动机器人与无线传感器网络(WSN)相结合,研究基于无线传感器网络的移动机器人导航系统,具有积极的科学价值和现实意义。针对传统遥控和无线控制作用距离有限的问题,本文设计并实现了无线传感器网络多跳通信链路,通过WSN自组网的功能,形成远大于传统方式的机器人控制链路,能够很好地满足复杂环境灾难救援的需求。同时,由无线传感器网络节点携带温度、湿度、气体浓度等传感器,实时感知周围环境的信息。机器人的机动性和强计算能力以及无线传感器网络的通信距离、分布式计算、环境感知等能力相结合,可以显著提高救援任务的执行效率。为增强移动机器人的运动控制能力,本文引入ROS机器人操作系统设计和构建了一种移动机器人平台,硬件采用了iRobot差分驱动底盘和UTM 30LX激光雷达。通过建立数学模型并进行分析,仅依托里程计的航迹推算方法具有较大的误差,而且随时间单调递增。因此,本文使用环境特征点对其进行矫正,通过扩展卡尔曼滤波的不断迭代,使得系统趋于稳定,误差协方差不随时间变化。在此基础上,针对室内环境,提出了基于Hector SLAM的同步定位与构图算法,使用激光雷达的数据匹配实现定位和局部地图到全局地图的连接。实验结果表明,所构建的移动机器人平台具有很好的鲁棒性,能够准确地构建室内环境的二维平面图,对于存在干扰的地面也能达到较高的精度。结合无线传感器网络全局动态感知的特点,本文提出了一种室内环境的移动机器人导航方法。将具备超声波传感器的WSN节点部署于天花板,通过预设的网格形式进行均匀放置。通过WSN节点感知其正下方的环境信息,在数据离散化后得到占有栅格地图,获得了动态环境的实时模型。在此基础上,本文引入D*Lite算法和PRM算法实施动态环境下的移动机器人路径规划。仿真实验表明,D*Lite算法能够在目标区域出现动态障碍时进行重规划,并显著减小对于包含动态障碍物的环境的路径规划时间。更多还原

【Abstract】 Since rescue robots can replace people to get to dangerous fields, such as, fires and mines, it attracts broad attention from domestic and overseas research institutes. Traditional rescue robots, limited by the short control distance, small range of perception and inaccurate positioning accuracy, had great difficulties in their application. This thesis, combing mobile robots and wireless sensor networks (WSN), studies mobile robot navigation system based on wireless sensor network. The research has positive practical significance and scientific value.This thesis designs and implements multi-hop wireless sensor network communication links to tackle traditional short range communication problem. Through the WSN’s self-construction function, it can form a much larger multi-hop communication chain, which can meet the needs of the rescue tasks of complex environments. At the same time, the wireless sensor network can carry some sensors, such as temperature, humidity, gas concentration, and can perceive the real time environment information, for the operators to plan the rescue mission. Mobility, robot’s strong computation ability and the communication distance, distributed computation of wireless sensor network combined can significantly improve the efficiency of executing a rescue mission.To enhance the capabilities of the mobile robot motion control, this thesis introduces ROS (Robot Operating System) design and build a mobile robot platform. Hardware uses the iRobot differential drive chassis and UTM 30LX laser radar. As using only dead reckoning odometer-based approach in mathematical models and analysis can lead a large error and the error covariance monotonically, therefore, environmental features to correct the position of robots is used, by continuously iterative using Extend Kalman filter to stabilize the system and error covariance does not change by time any more. On this basis of mapping, this thesis proposed a simultaneous location and mapping algorithm based on Hector SLAM. Laser data matching system is used to locate the robot. Local map is connected to a global map. Experimental results show that the algorithm is relatively robust and can achieve high accuracy in rough terrain.For indoor mobile robot navigation, an algorithm based on wireless sensor network is proposed in the thesis. The WSN nodes with ultrasonic sensors are deployed on the ceiling, by predefined form of the grid. The WSN nodes can perceive the real-time environment information under it, then occupancy grid map is constructed by data discretization, then the dynamic model is get. This Thesis introduces D*Lite algorithm and PRM algorithm to plan a shorted path for mobile robot in dynamic environments. Simulation results show that D*Lite algorithm can re-plan, and significantly reduce dynamic path planning time in the environment filled with obstacles.更多还原