Robot environment sensing method based on multi-sensor fusion

By combining geometric primitive layer-by-layer comparison with local factor graph optimizer, high-precision matching and pose correction of multi-sensor data are achieved, solving the problems of insufficient matching accuracy of multi-sensor data and map update lag in dynamic environments in the existing technology, and improving the robot's navigation ability in complex environments.

CN122468083APending Publication Date: 2026-07-28YOUDOU (BEIJING) TECHNOLOGY CO LTD
0 Cites 0 Cited by

Patent Information

Application Number
CN202610606108.8
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2026-05-06
Publication Date
2026-07-28

AI Technical Summary

Technical Problem

Existing technologies lack semantic landmark matching mechanisms based on layer-by-layer comparison of geometric primitives, resulting in insufficient matching accuracy between multi-sensor data and topological maps. The pose correction process does not comprehensively consider semantic categories, observation times, and keyframe conditions, making it difficult to adapt to the multi-source data fusion requirements in complex dynamic environments. Furthermore, there is a lack of effective processing and map updating methods for dynamic obstacle areas.

Method used

By comparing geometric primitives layer by layer, it is determined whether there are semantic landmarks in the topology map that match the original multi-sensor data. The set of matching landmarks is selected for pose correction. When there are no matching landmarks, the grid map interface is displayed for users to set dynamic obstacle areas. The topology map is constructed by combining the local factor graph optimizer for joint initialization and bundle adjustment optimization.

Benefits of technology

It improves the accuracy and robustness of robot pose estimation in complex environments, enhances its adaptability to dynamic environments, and solves the problems of insufficient matching accuracy and lagging map updates.

✦ Generated by Eureka AI based on patent content.
Patent Text Reader

Abstract

The invention provides a robot environment sensing method based on multi-sensor fusion. The robot environment sensing method comprises the following steps: performing time-space synchronization processing on original multi-sensor data to generate synchronous sensing data; judging whether semantic landmarks matched with the original or synchronous data exist in the topological map or not, wherein the matching condition is that geometric primitives of the semantic landmarks in the topological map are the same as geometric primitives in the data; when determining that the matched landmarks exist, selecting the matched landmarks to form a set; and performing pose correction on the original multi-sensor data based on the set, fusing different pose constraints in the matched landmarks into the original data, and obtaining corrected data for robot path planning. According to the method, the consistency of multi-sensor data and the robot environment sensing precision can be improved, and reliable correction data is provided for path planning.
Need to check novelty before this filing date? Find Prior Art