基于深度信息的移动机器人室内环境三维地图创建

针对使用扩展卡尔曼滤波(EKF）进行环境地图的创建对线性系统效果较好而对非线性系统的线性化受误差影响较大的问题,提出一种基于对Kinect采集到的环境数据和迭代扩展卡尔曼滤波(IEKF）算法的室内环境三维地图创建。该方法使用成本较低的Kinect传感器获取深度数据然后结合IEKF实现摄像头轨迹预测,最后利用最近点迭代(ICP）算法对深度图像进行配准得到室内环境三维点云图。实验结果表明,IEKF算法与传统的EKF算法相比,得到的轨迹更平滑、误差更小,同时所得到的三维点云图更加光滑。该方法实现了三维地图构建,较为实用,效果较好。

Depth-image based 3D map reconstruction of indoor environment for mobile robots

Considering the problem that Extended Kalman Filter (EKF) does better in linear system for real-time 3D mapping and largerly affected by errors to linearize nonlinear systems, Iterated Extended Kalman Filter (IEKF) based on depth data of Kinect was proposed. This method used IEKF to achieve camera trajectory prediction applied to Microsoft Kinect RGB-D(Red-Green-blue-Depth) data, after that Iterative Closest Point (ICP) algorithm was employed to perform fine registration on depth image to generate the 3D point cloud map. The experimental results show that compared with the traditional EKF algorithm, the IEKF generates less error than EKF, and gets the more smooth 3D point cloud map. The method realizes the 3D map-building, and it is more practical.