Scalable and flexible vision-based multi-robot tracking system

Abstract

Precise position is necessary for a robot to reliably localize itself and navigate in the environment. Robot position can be calculated internally or acquired from an external system. This paper presents a software architecture of a vision-based robot tracking system that can provide precise position information as well as identification of many robots. The software architecture has been designed to take into consideration the need for a flexible and scalable system that can interoperate with many similar systems and support different types of image sources, image processing, and outputs. With the architecture, future enhancements or replacements can be implemented as plug-ins, which can be easily integrated into the system. This paper also demonstrates the proposed architecture in different application scenarios.