基于改进扩展卡尔曼滤波算法的移动机器人定位方法研究

Research on Positioning Method of Mobile Robot Based on Improved Extended Kalman Filtering Algorithm

针对移动机器人在位姿跟踪过程中存在单一传感器或多传感器测量系统对环境信息处理能力有限的问题,结合扩展卡尔曼滤波算法,对传感器测量信息进行融合分析.对于单个传感器测得的n个观测值,扩展观测矩阵至大于n的m个目标测量值,将预测空间到测量空间的映射设计为一个具有n个非零变量、维数为nm、秩为n的变换矩阵,实现传感器对状态向量的局部更新.在建立的传感器及机器人运动模型基础上,通过地面移动机器人进行实验验证.理论分析和实验结果表明,该方法能在保证定位精度的前提下,提高算法对不同传感器类型和传感器数量的泛化能力,增强测量系统的准确性和灵活性.

Aiming at the problem that single sensor or multi-sensor measurement system has limited ability to process environmental information in the process of position and orientation tracking of mobile robots,the author made the fusion analysis on the measurement information combine with the extended Kalman filter algorithm.For nobservations measured by a single sensor,the author extended the observation matrix to m target measurements and designed the mapping from the prediction space to the measurement space to a transformation matrix which has nnon-zero variables,nmdimension and rank nto achieve local updating of the state vector by the sensor.Experiments were taken on the ground mobile robot based on the established sensors and robot motion mathematical model.Theoretical analysis and experimental results show that the proposed method improves the generalization ability of the algorithm for different types and numbers of sensors under the premise of ensuring the positioning accuracy,meanwhile,enhance the accuracy and flexibility of the measurement system.