融合激光测距仪和惯导信息的移动机器人室内定位方法研究

发布时间：2018-08-21 20:45

【摘要】：随着工业机器人技术的成熟,智能移动机器人进入蓬勃发展的阶段,其中定位技术是解决移动机器人导航,实现全自主式移动机器人的关键技术。相比于室外环境下的移动机器人定位,室内环境下的定位精度要求更高,也更难实现。本文以双轮差动式室内移动机器人为平台,以激光测距仪、微惯性测量单元和编码器为主要传感器,探索在已知结构化环境下的移动机器人室内定位方案,分别从理论和实践上进行了研究。首先,分别基于里程计和微惯性测量单元建立了两轮差动式室内移动机器人的运动学模型;并就室内地面不平整或倾斜的情况提出了一种用于计算机器人初始姿态的对准算法;利用卡尔曼滤波将里程计和惯导数据进行融合,将运动过程中的横滚角和俯仰角信息引入到位置更新算法中,有助于减小因地面坑洼、凸起或车轮打滑而导致的定位误差;最后分别基于里程计、微惯性单元以及里程计/微惯性组合进行了仿真,验证了所提出的位置更新算法的正确性。其次,针对激光测距仪数据线段分割中的过分割和过合并问题提出了一种基于Split-Merge框架的线段分割策略,并使用Hough变换方法拟合圆弧特征,成功提取出了环境中的线段和圆弧特征。然后,在环境特征提取的基础上,制定了室内移动机器人初始定位方案和动态定位方案。初始定位算法以完整线段为特征进行地图匹配,采用微惯性器件辅助激光测距仪的方式进行定位,优化后的定位算法在非相似和对称相似环境下进行仿真均取得了较好的定位效果。动态定位方案采用紧耦合的方式,使用无迹卡尔曼滤波对作为组合定位滤波器,减小了计算量,提高了定位精度,仿真结果表明了所使用的组合定位方案的有效性。最后,针对本文研究的定位方法搭建了室内移动机器人实验平台,分别在非相似环境和对称相似环境两种结构化环境下进行实验研究。结果表明,室内移动机器人在两种环境下均取得了较好的定位效果,特征提取算法能够正确拟合环境中的线段和圆弧特征,初始对准算法的加入使激光测距仪在对称相似环境下能够正确辨识航向角,组合定位方案可以正确估计出机器人初始位置和运动过程中的轨迹。
[Abstract]:With the maturity of industrial robot technology, intelligent mobile robot has entered the stage of vigorous development, among which positioning technology is the key technology to solve the navigation of mobile robot and realize the full autonomous mobile robot. Compared with the mobile robot positioning in outdoor environment, the positioning accuracy in indoor environment is higher, and it is more difficult to achieve. Based on the two-wheel differential indoor mobile robot and laser rangefinder, micro-inertial measurement unit and encoder as the main sensors, this paper explores the indoor positioning scheme of the mobile robot in the known structured environment. Respectively from the theory and practice of the study. Firstly, the kinematics model of the two-wheel differential indoor mobile robot is established based on the mileage meter and the micro-inertial measurement unit, and an alignment algorithm for calculating the initial attitude of the robot is proposed to calculate the initial attitude of the robot when the indoor ground is uneven or tilted. Kalman filter is used to fuse the odometer and inertial navigation data, and the information of roll angle and pitch angle in the course of motion is introduced into the position updating algorithm, which is helpful to reduce the positioning error caused by the ground potholes, bumps or wheel skidding. Finally, based on the mileage meter, the micro inertial unit and the mileage / micro inertial combination, the correctness of the proposed position updating algorithm is verified. Secondly, aiming at the problem of over-segmentation and over-merging in data line segment segmentation of laser rangefinder, a line segment segmentation strategy based on Split-Merge framework is proposed, and Hough transform method is used to fit the arc feature. The features of lines and arcs in the environment are extracted successfully. Then, based on the extraction of environmental features, the initial localization scheme and dynamic localization scheme of indoor mobile robot are established. The initial location algorithm uses the complete line segment as the feature of map matching, and uses the micro-inertial device assisted laser rangefinder to locate. The simulation results of the optimized localization algorithm in dissimilar and symmetrical similar environments are satisfactory. The dynamic location scheme adopts the tight coupling method and uses the unscented Kalman filter as the combined positioning filter. The computation is reduced and the positioning accuracy is improved. The simulation results show the effectiveness of the combined positioning scheme. Finally, the indoor mobile robot experimental platform is built for the localization method studied in this paper, and the experimental research is carried out in two kinds of structured environments: dissimilar environment and symmetrical similar environment. The results show that the indoor mobile robot has achieved good localization effect in both environments, and the feature extraction algorithm can fit the line segment and arc feature of the environment correctly. With the addition of the initial alignment algorithm, the laser rangefinder can correctly identify the heading angle in the symmetrical similar environment, and the combined positioning scheme can correctly estimate the initial position and the trajectory of the robot in the course of motion.
【学位授予单位】：哈尔滨工业大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP242

【参考文献】