Mobile Agent-based Remote Vision Sensor Fusion

Abstract

Vision systems have become popular for remote vision sensing in geographically distributed environments due to vast amount of information they provide. However, remote vision sensors are generally plagued with power and communication bandwidth constraints. Mobile agent technology is a salient solution to geographically distributed and dynamic domains that require subsystems to interact with each other. Mobile agent technology increases power efficiency by reducing communication requirements and increases fusion processing by allowing in-situ integration of on-demand visual processing and analysis algorithms. A mobile agent can dynamically migrate from one vision sensor to another and combine all necessary sensor data in a desired manner specific to the system requesting the data. This paper presents a remote vision fusion architecture based on mobile agent technology that provides a flexible vision fusion solution. The architecture utilizes the Mobile-C library as a basis for the mobile agency along with OpenCV and ImageMagick for vision processing and manipulation. An application example is provided that demonstrates the benefit of using mobile agents for vision fusion in remote vision systems. The validity of the architecture is proven in an experimental setup with a retrofitted robotic cell comprised of a Puma 560, IBM 7575, a conveyor system, and a vision system.