Thursday, February 28, 2008
Developing new machine-vision software applications
Christian Demant, NeuroCheck and Industrial Vision Systems, on the demands of advanced development platforms
The fast development cycles in the IT industry result in permanent pressure on machine-vision software developers to update their knowledge. Integral parts of that continuous learning process are new networked system architectures and software-development tools.
Advanced new development platforms like .NET provide the option to easily combine software modules written in several different programming languages. This supports the development in teams having different professional backgrounds. Therefore, developers managing these teams must have knowledge in all these programming languages.
The availability of the latest multicore CPU technology demands multithreaded software to take advantage of this new processing power. Enabling software for parallel computation adds a totally new level of complexity to the development process. The normal programming approach, being used for dozens of years, is now, in many cases, obsolete. The synchronization of parallel executing threads leads to new error situations, which are extremely difficult to predict and to simulate in advance.
In the future I see growing importance of an in-depth understanding of software design patterns. The growing size and complexity of machine-vision software applications requires a much more systematic approach during the design and planning stage. Even 10 years ago a software developer started the implementation a couple of hours after discussing the specification with his sales department.
The software-development process has moved toward something comparable to a team of architects in charge of a complex facility. The design of the building and the drafting of the plans and specifications are the main intellectual tasks and take a big part of the development time. The implementation afterward is handcraft, but both jobs together require skilled teams. Lone fighters have no chance of survival.
Christian Demant is general manager, NeuroCheck, Stuttgart, Germany, and
Director, Industrial Vision Systems, Kingston Bagpuize, UK
Subscribe to:
Posts (Atom)