Machine Vision for intelligent Semi-Autonomous Transport (MV-iSAT)

The primary focus was to develop a vision-based system suitable for the navigation and mapping of an indoor, single-floor environment. Devices incorporating an iSAT system could be used as ‘self-propelled’ shopping carts in high-end retail stores or as automated luggage routing systems in airports. The primary design feature of this system is its Field Programmable Gate Array (FPGA) core, chosen for its strengths in parallelism and pipelining. Image processing has been successfully demonstrated in real-time using FPGA hardware. Remote feedback and monitoring was broadcasted to a host computer via a local area network. Deadlines as short as 40 ns have been met by a custom built memory-based arbitration scheme. It is hoped that the iSAT platform will provide the basis for future work on advanced FPGA-based machine-vision algorithms for mobile robotics.