A Survey on Mobile Robot Visual Relocalization in Complex Dynamic Scenes

Read full text Summarise with VIRAT
Authors: Shichu Sun
Cited by: 0
References: 13

Summary

No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.

Summarise with VIRAT Free to try with your Science Catalogs account.

Abstract

Visual relocalization is key to enabling continuous autonomous navigation of mobile robots. This survey systematically reviews research progress on visual relocalization technology in complex dynamic scenes, focusing on the issues of excessive computational load and mismatches caused by dynamic disturbances in real environments. It explores the evolutionary trajectory from global representations and local matching to semantic learning methods, revealing that the core mechanisms for handling dynamic disturbances have shifted from passive feature filtering to "active suppression" using spatial attention, adaptive matching, and soft weighting. Taking into account the deployment requirements of mobile robots, the paper summarizes engineering solutions that encompass lightweight matching and multimodal priors, and envisions future research directions such as deep semantic-geometric collaboration and incremental map evolution. These insights provide theoretical reference and practical guidance for the design of robust localization systems.

Fields of Study & Subjects

Robotics and Sensor-Based Localization Advanced Image and Video Retrieval Techniques Advanced Vision and Imaging

Publication details

Journal / source
Advances in Computer Signals and Systems
Volume
10
Issue
1
Publication date
1 January 2026
Language
EN