Abstract:Aiming at the problem that the indoor mobile robot cannot accurately estimate the pose in a dynamic environment, a RGB_〖KG-*2〗D visual synchronous positioning and mapping method combining frame difference method and background compensation was proposed. The ORB method was used for feature point detection and matching, then the LK optical flow method was used to partially remove the impact of dynamic objects on global motion estimation and to perform background compensation.Finally, the frame difference method was used to obtain the dynamic target, and the dynamic target was removed from pose estimation.The feature points, coupled with morphological erosion and dilation operations, were used to remove the effects of noise on the target extraction process. Experiments on the TUM dataset show that the proposed method can effectively improve the positioning accuracy and reduce the impact of dynamic objects.