Loop closure detection of visual SLAM based on variational autoencoder

Shibin Song, Fengjie Yu, Xiaojie Jiang, Jie Zhu, Weihao Cheng, Xiao Fang · 2024

Read Loop closure detection of visual SLAM based on variational autoencoder by Shibin Song, Fengjie Yu, Xiaojie Jiang, Jie Zhu, Weihao Cheng, Xiao Fang (2024) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/pmc-10837850.txt.

Description

Song and co-authors improve how mapping robots recognize places they have seen before, in a study published in 2023. The task, loop closure detection, corrects the drift that builds up as a robot maps a space. Older methods rely on hand-made image features, which can misfire when lighting or scenes change. The team use a variational autoencoder, a neural network that compresses each image into a compact code.

An attention mechanism and a tuned loss sharpen the image codes. A geometric check then weeds out false matches. In experiments, the method beat the classic bag-of-words approach and other deep learning methods on precision and recall. Tests across three kinds of scenes showed it holds up in real-world use.

Catalog record

CREATORShibin Song, Fengjie Yu, Xiaojie Jiang, Jie Zhu, Weihao Cheng, Xiao Fang
YEAR2024
COLLECTIONPAPERS
FORMATTXT
FILE SIZE30 KB
VERSION1.0
LICENSECC BY 4.0
RIGHTSCC BY
SOURCEEurope PMC

Source: https://europepmc.org/articles/PMC10837850

Browse the full library · About the Tower of Babel