【作者】 丁毅；

【导师】 邱白晶；

【作者基本信息】 江苏大学 ， 机械设计及理论， 2005， 硕士

【摘要】 机器人已经广泛应用于工业、农业和军事等诸多领域。本课题中的割草机器人采用电动履带式机器人作为研究平台,具有其他割草机器人所没有的地形适应性强和全天候工作的优势,为割草机器人平台的选择提供了新思路。 割草机器人研究的主要内容之一就是如何实现机器人平台的导航定位,目前大部分机器人采用视觉、声学等传感器提供导航信息,而已有的割草机器人大多使用电缆围出工作地域、电涡流传感器获得工作范围信息的方式进行导航。本课题使用GPS以及GPS/DR组合导航方式用于割草机器人的导航定位,建立的GPS卡尔曼滤波模型和GPS/DR组合导航的非线性自适应卡尔曼滤波模型以提高机器人导航定位的精度,经过试验验证取得了比较好的效果。 本文首先对GPS的误差是否是有色噪声进行了研究,利用数值分析和系统辨识的方法对AgGPS132型接收机工作于伪距差分模式所采集的定点静态定位数据进行处理,建立了GPS误差的自回归模型—AR(1)型。GPS卡尔曼滤波器中都把GPS误差作为高斯白噪声来处理,本文利用建立的GPS误差模型对GPS卡尔曼滤波器进行改进,得到有色噪声下的GPS卡尔曼滤波模型,这一模型经过试验检验,与单一GPS动态定位比较,定位精度显著提高。 为了验证GPS动态定位在不同的行进方向下是否存在差异,利用机器人和GPS接收机进行东西和南北方向的动态定位试验,得到了东西方向动态定位精度好于南北方向的动态定位精度这一结论,这一结论对利用GPS进行导航定位的机器人的路径规划具有指导意义。 为了进一步提高GPS定位精度,采用GPS接收机、数字罗盘和机器人主动轮编码器读数换算得到的里程信息组成GPS/DR组合导航系统,并且建立了GPS/DR组合导航的卡尔曼滤波模型,通过试验检验,GPS/DR组合导航方式下的定位精度进一步提高。 文章最后比较了GPS动态定位、经过卡尔曼滤波的GPS动态定位、DR航位推算法和GPS/DR组合导航定位这四种方式下的定位精度,分析误差来源,并且提出了进一步改进的意见。本课题中建立的两种卡尔曼滤波模型显著提高了GPS定位精度,对实现机器人导航定位具有一定的应用价值。更多还原

【Abstract】 The robot has widely applied in the fields of industry、 agriculture and military affairs.The mowing robot in this paper used electric pedrail robot as a research platform for the first time, and the electric pedrail robot has the advantage of the landform applicability and all-weather that other robot has not. This choise has created a new method in the research of the mowing robot.One of the most important research problems on the mowing robot is how to carry out the aim at navigation and position. At present, most of robots use the sensor of vision or the sensor of sound to obtain the navigation information. Other mowing robots use the current transducer to obtain the information of its work area from the cable that laied at work area beforehand. In this paper, we headmostly apply the GPS and GPS/DR navigation technic in the mowing robot and establish the GPS Kalman filter model and GPS/DR adaptive Kalman filter model. These models have good effect by test.First,this paper introduced the research on GPS error, we aim at examing the error of GPS is the colour noise or not.We collected position data on a stationary site by using AgGPS 132 receiver. By using the methods of numerical analysis and system identification, the Arma model [AR(1)] of GPS error was establishedin GPS Kalman filter,the GPS error is treated as Gauss white noise commonly.But we use the GPS error model which established to improve the GPS Kalman filter and establish the GPS Kalman filter model in the condition that GPS error is colour noise. After experiment testing,the GPS Kalman filter model has inproved position precision greatly than GPS.In order to validate that the GPS dynamic navigation error is diversity in different direction, we use the AgGPS 132 receiver in dynamic navigation in direction of east -west and north-south,then we proved that the error of GPS’s dynamic navigation in different direction is distinguishing.This result has directive significance in robot’s route plan.We use the GPS receiver、 the digital box and needle and the distance information calculating from robot’s PWM assembled the GPS/DR navigation system and establish the GPS/DR adaptive Kalman filter model.The GPS/DR adaptive Kalman filter model has improved position precision more by experiment testing.Finally, this paper analyzed the source of position error and compare the results of examination among four navigation method, that is the GPS dynamic position method,GPS Kalman filter method,DR method and GPS/DR adaptive Kalman filter,and give some advice in mending.The GPS Kalman filter model and the GPS/DR adaptive Kalman filter model in this paper improve the position precision greatly and they are worth of application in robot navigation and position.更多还原