Skip to Main Content
Automation traditionally has involved the use of sensors that simply transfer raw data signals to programmable logic controllers (PLCs). With advent of smart cameras and intelligent terminals, distributed computing has enabled decentralized control and monitoring of sensors. This, has in turn, enabled us to conceptualize a wide range of network configurations that derive from centralized and decentralized control elements. Any hybrid configuration can be analyzed as a combination of these two basic configurations. To arrive at an optimal configuration, comparison of architectures and their parameters such as the number of subsystems, cameras, provisions for refinement and scalability of throughput is needed. Such an analysis is useful in conveyor based machine vision applications such as package inspection where response times are critical, and deterministic. In this paper, a generic tree structure for an automation system is presented and performance characteristics for these configurations are analyzed by tabulating three critical quantities involving execution time of an embedded program, latencies of I/Os and transfer of data to the process image in PLC. From the performance times, it is evident that while centralized control exhibits ease of maintenance and troubleshooting, decentralized control demonstrates better speed, smaller memory footprint and resilience to failure of subsystems. Configurations suitable for different machine vision applications are also examined. Examples of various conveyor based automation solutions implemented in the factory floor using SIEMENS PLC, S7400 and Siemens smart cameras, SIMATIC VS7xx substantiate the results.