Robust, Component-Based Vision Software Architecture For Future Combat Systems

While machine vision is a critical enabler for Intelligent Automation, technology innovation has been slowed by custom development and high integration costs. This proposal addresses these issues by advancing a modular, component-based design that allowsvision integrators to customize applications from prefabricated, interchangeable parts. The component-based approach provides end-user flexibility, while affording large economies through software reuse. In particular, the project will define aReference Architecture for component-based vision systems. This definition includes application specifications, as well as functional and syntax requirements for all components and interfaces within the domain. The proof of concept is demonstrated byapplying the proposed Reference Architecture design to a visual servo application. Specifically, visual servoing provides visual feedback cues for servo-level control. The primary benefit being that vision is used directly in the control loop, whichprovides greater automation capabilities and better overall performance. The vision system will be integrated into the Smart Crane Ammunition Transfer System (SCATS) to allow operator control of the pallet loading/unloading process from inside the cabin.The subsequent follow-on effort will demonstrate software reuse by reconfiguring the vision system for different hardware platforms and for different missions. The success of this project will directly impact both developers and end-users of visionsystems. From the development perspective, the component and interface specifications will remove the proprietary barriers that many developers erect in an attempt to protect intellectual property. This will encourage greater competition and fosterinnovation at a greater level of granularity. To the vision integrator, component-based standards will provide the flexibility to choose the best implementation of a subsystem for his/her particular needs. Additionally, componentization will bring downthe overall price of vision systems, making it more practical and accessible to a wider range of audiences.