A real-time mapping method based on sparse-slam

Abstract

The present invention provides a sparse-slam-based real-time mapping method, including the following two threads running in parallel: thread 1, tracking and extracting key frames; thread 2, creating and inserting point cloud MAP. The real-time mapping method of the present invention adopts two parallel threads, which can ensure the real-time performance of the two threads, which not only saves time, but also preserves the characteristics of the image as much as possible. On the premise of ensuring efficient and real-time algorithm, more scene information.

Description

technical field

[0001] The invention relates to a real-time mapping method, in particular to a sparse-slam-based real-time mapping method. Background technique

[0002] SLAM technology has become known with the popularity of robots, VR, and AR in recent years, and has made different progress in sensors, algorithms, software, and hardware. SLAM (simultaneous localization and mapping) is divided into two major functions, positioning and mapping. It mainly solves the problem that when a device (such as a robot, VR device, etc.) comes to a completely unfamiliar environment, it needs to accurately establish the corresponding relationship between time and space, and can perfectly answer the following series of questions: Where was I just now, where are you now? What did I see, and how is it different from what I saw before? What is my past walking trajectory? What does the world I see now look like, and how has it changed from the past? Is my trajectory shaking, is my positio...

Images