Navigation method based on depth camera/MEMS inertial navigation/odometer combination

The invention provides a navigation method based on a depth camera/MEMS inertial navigation/odometer combination. The method comprises the following steps: combining a depth camera, MEMS inertial navigation and a speedometer by adopting a loose combination mode, reasonably selecting images, and adopting a Kalman filtering mode based on a position difference value, so the navigation positioning error of the intelligent mobile robot in an indoor scene is reduced, and an indoor three-dimensional reconstruction model is obtained. According to the method, the visual navigation precision is improved while the problems of short-term high precision and long-term error divergence of inertial navigation are solved. 本发明提出一种基于深度相机/MEMS惯导/里程计组合的导航方法，采用了松组合的方式将深度相机、MEMS惯导与里程计进行组合，通过合理的选取图像，采用基于位置差值的卡尔曼滤波方式，减小了室内场景下智能移动机器人的导航定位误差，并得到室内的三维重构模型；该方法在解决惯导导航短期高精度、长期误差发散问题的同时，提高了视觉导航精度。