基于Kinect和惯导的组合室内定位

摘要

针对机器人在室内移动导航时精度不高和误差积累等定位问题，研究基于 Kinect 传感器和惯导的组合定位方法，引用一种适合于数字近景影像的绝对定向算法计算机器人在相邻位置上的姿态信息和平移量，提高机器人的移动参数的计算精度。在Kinect视觉测程基础上进行惯导器件数据的捷联解算，利用卡尔曼滤波将两者输出的位置和姿态的误差作为观测量进行最优估计以修正误差。实验结果表明，该方法能够提高室内移动机器人的定位精度和稳定性。%Indoor navigation has been suffering from the low location accuracy and accumulation error,which is investigated with the Kinect-based visual location aided with initial devices to improve the deficiency.Absolute orientation algorithm was used to calculate the robot’s profile information and translation on adj acent position,thus improving the calculation precision of the moving parameters of the robot.Kinect was advised as the sensor of position analysis and the inertial computation as another solution,and both were incorporated into the Kalman filtering to find the optimum estimation from individual outcomes and atti-tude errors.Experimental results demonstrate that the proposed method can improve the location accuracy and stability of indoor robot navigation literally.