Semantic information and vslam fusion method for sweeping robot

Abstract

The present invention provides a semantic information and VSLAM fusion method for sweeping robots. The method adds the vector containing voice information in the semantic dictionary to the front of the traditional dictionary vector to generate a fusion dictionary that fuses traditional information and semantic information, and strengthens VSLAM. The information source of the system has changed the shortcoming that the traditional VSLAM cannot obtain the prior information of the environment, and uses the semantic information to improve the accuracy of the VSLAM system to solve the essential matrix; in the loop detection, the semantic information is first matched. If the semantic information cannot complete the matching, then If the point is considered to be a wrong match, there is no need to search in the bag of words, which improves the robustness of the system and the accuracy of building an indoor map.

Description

technical field [0001] The invention belongs to the technical field of synchronous positioning and mapping (SLAM), and in particular relates to a fusion method of semantic information and VSLAM for a sweeping robot. Background technique [0002] Sweeping robots are becoming more and more common in people's lives. The core technology of the sweeping robot includes sweeping, mopping, obstacle avoidance, mapping and human-computer interaction. Among them, in addition to sweeping the floor, there are varying degrees of problems in functions such as mopping the floor, and they are still in the exploratory stage. [0003] Position and map are often closely related, that is, positioning and map are interdependent, and positioning cannot be discussed without a map. However, if you want to build a map suitable for robots through indoor robots, you need to know the position of the robot itself. Since there are often items such as tables, chairs, boxes and cabinets in the room or the...

AI Technical Summary

Problems solved by technology

The positioning and mapping of indoor robots is mainly realized through SLAM technology. The mainstream solutions of SLAM technology include visual SLAM and laser SLAM. Visual SLAM is a current research hotspot, which has the advantages of low cost and rich information, but the stability and accuracy of visual SLAM are relatively low. Poor, and more complex than laser SLAM; laser SLAM has better results in indoor robot positioning and mapping, but the laser data is single, and it cannot achieve closed-loop detection well, and the low-cost laser radar has a low laser point density. At the same time, there is an occlusion phenomenon, so the constructed map often cannot be closed; and the semantic information of the picture can provide more information for SLAM, and provide more accurate semantic information in the subsequent process, so the combination of semantic information and SLAM is the best choice. a trend

Images