.png){kind=icon}
.png){kind=icon}
Return
Collaborative Visual Localization for Modular Self-Reconfigurable Robots
.png){kind=icon}
OA
AI
DOI：10.1002/aisy.202501291.png)
.png){kind=icon}
.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
Relative localization remains a significant challenge for modular self-reconfigurable robots (MSRRs), particularly due to their high integration and limited onboard hardware, which often lead to constrained fields of view, sensor faults, and intermittent observations. In response to these common issues in MSRR systems, this paper presents a vision-based collaborative localization approach developed through a hardware–software codesign framework, implemented and tested on the SnailBot platform. Each modular unit is equipped with a monocular camera and a lightweight processing unit, while multiple ArUco marker arrays are distributed across the spherical shell to support embedded visual perception and facilitate relative pose estimation across a variety of viewing angles. At the algorithmic level, we introduce a learning-based cooperative localization method that includes fault tolerance and real-time anomaly detection, aimed at mitigating the impact of sensor failures and occasional perception loss. The approach is designed to work with sporadic and asynchronous visual observations, allowing for state recalculation and alignment among modules without relying on continuous measurement. In our experiments, the proposed method shows encouraging results in terms of accuracy, robustness, and consistency under several fault and occlusion scenarios, including outdoor tests where the localization system was deployed and evaluated on the robot in unstructured environments. These results suggest its potential as a practical solution to support perception and self-reconfiguration in MSRR applications.
Keywords:
collaborative localization
modular self-reconfigurable robots
multi-robot systems
AI Summary
Key information extracted from the uploaded paper, including a brief overview, abstract, background, key highlights, visual analysis, and future outlook.
Journal
IF:
6.1
Papers:
2.0K
Citations:
8.4K