Computing embedded platform for transportation, industry and surveillance

To create a new embedded computing platform for real time data processing needs combining stadard cpus and programmable hardware as well as tools for efficient development of software and testing. The platform should be smart, energy efficient, and economically feasible.

Video and signal processing are important fields of indutrial, traffic, and surveillance applications. They are also in focus of the proposed project participants - CAMEA, FIT BUT, goHDR and RTS. The applications of video and signal processing include the traffic monitoring and enforcement, such as section speed meaasurement, red light enforcement, weighing in motion, and many other ones. Another field of application of such systems is the industry, the main field of research is production quality inspection, such as quality control of electronic components or nonwoven textiles. Video processing is also impoerant in surveillance applications, such as city video camera systems. These applications and reseach fields (the participants of the proposed project are also active in research projects) are all covered by the proposed project participants and they already do have hundreds of systems in video traffic monitoring and surveillance, weigh in motion, as well as quality inspection, in place and these systems do already have significant societal and economical impact. One of the main problems of development of applications in computer technology applications is the need for real-time data processing, where the image and video processing, measurements, or processing of other sensory signal inputs (radars, induction loops, weighing sensors, etc.). The typical platform being used in these applications is a personal computer (PC). In applications that are demanding from the computational point of view and in which one computer (or CPU) is not enough, parallel computational approach is often used, where more computers (CPUs) are being used in parallel. In such situations, in order to achieve the needed speedup, the CPUs must be suitably connected and the tasks decomposed. Contemporary technology offers also multi-core CPUs or graphics processors (GPUs). The problem of these devices in real applications ofte is that they consume unacceptably high amount of energy, they are relatively large and expensive as well. The computational devices to be used must be economically feasible, they must be low power devices, and they must have enough computational power. Given the current state of the art, heterogeneous computational platforms, containing different computational resources, are a good solution where the exploitation of their individual parts is optimized. The computationally most demanding parts of the aplpications typically run in specialized digital circuits or specialized processors (such as digital signal processor - DSP) while the rest runs on a standard processor (CPU). Thanks to the fact that the computationally expensive part runs (accelerated) in hardware, the total speed of execution of the algorithm can even be faster than in the traditional CPUs (in PC). The accelerated part can be implemented in (ASIC - Application Specific Digital Circuit); however, usage of ASIC is limited as they need to be manufactured specifically for each task and this is very expensive so that it is economical only in large series production and therefore for large manufacturers. For this reason, technologies of programmable hardware - mostly gate arrays (FPGA - Field Programmable Gate Arrays) are often used as they offer high computational speed, massive parallelism, and also good flexibility with good price. During development of the computational systems, it is necessary to take into account economy not only the economy of their productio but also, perhaps more importantly, the efficiency and economy of the software and application development process (programming languages, compilers, operation systems, libraries of functions) is oftern more critical than the price of the equipment itself. The reason for this is that the price of human labour. The goal of the project is to create a novel computational platform that fulfills the requirements of the modern applications, presumably based on technologies, such as Xilinx Zynq with FPGA and CPU combined with a software/firmware development framework supporting software in C, C++, and C# languages and compatible with PC. After a careful discussion of the partners, it has revealed that combining their fields of expertise, such a development can be successfull. CAMEA will use its expertise in design, manufacturing, and applications in traffic and industry, FIT will use its expertise in algorithm research, RTS will use its signal processing skills, and goHDR its exoertise in HDR and surveillance. The project will be subdivided into several phases focusing individual parts of the development - architecture, methods of application development, hardware of the platform, experiments, and validation. The output will be mainly software, functional samples (demonstrators of the individual applications), and prototpe of the platform.
Project ID: 
9 550
Start date: 
Project Duration: 
Project costs: 
920 000.00€
Technological Area: 
Data Processing / Data Interchange, Middleware
Market Area: 
Data processing, analysis and input services

Raising the productivity and competitiveness of European businesses through technology. Boosting national economies on the international market, and strengthening the basis for sustainable prosperity and employment.