摘要
提出了一种基于车辆运动微分模型的扩展卡尔曼同步定位与地图创建(EKFSLAM)算法,该算法是将车辆运动轨迹视为由许多微小直线段组成,利用车辆行驶环境中的柱状特征,结合扩展卡尔曼滤波器,实现移动机器人的同步定位与地图创建(SLAM).该算法能有效地降低单纯使用航迹推算所产生的定位误差,定位精度与基于运动学模型的EKF-SLAM在同一量级.与基于车辆运动学模型的EKFSLAM算法相比,该算法不仅具有更为简化和通用的模型表达形式,同时由于该算法所需的数据可以更方便地进行精确测量,不易受到噪声的干扰,因此稳定性有一定的提升.
Based on a vehicle-motion differential model, a simuhaneous localization and mapping ( SLAM ) algo- rithm hased on the extended Kalman filter (EKF-SLAM) is proposed in which the vehicle trajectm'y, whieh is considered to be composed of many small straight line segments, and the SLAM of a robot is realized using the columnar feature of the traffic environment and the EKF. The proposed algorithm can efficiently reduce the positioning error caused when only dead reckoning is used, i. e. , the positioning accuracy of the proposed al- gorithm is of the same order of magnitude as that eft the EKF-SLAM algorithm based on a vehicle kinematic model. Compared with the EKF-SLAM algorithm based on the vehicle kinematic model, the proposed algo- rithm has a simpler and more general model configuration. Meanwhile, because accurate measurement of the data required by the proposed algorithm can be done conveniently and thus less susceptible to noise at the sen- sor node, the stability of the system can be improved to some extent.
出处
《信息与控制》
CSCD
北大核心
2014年第1期82-87,共6页
Information and Control
基金
国家自然科学基金资助项目(91120307
50908222)
江苏省自然科学基金资助项目(BK2012151)