About the innovator

Since 1996 Haiku Tech has been providing complete production lines, material consultancy, process optimization and upscaling services for developers and producers of multilayer ceramics. Energy conversion and storage, passive electronic components, piezo actuators, LED lighting, and printed circuitry are a few examples of the industries which benefit from our services. Haiku Tech deploys its broad experience and deep understanding of ceramic systems to develop optimal technological solutions for its customers, in order to reduce development cycle times and limit the cost of upscaling. We are keen to take on challenging technological projects and push the boundaries of technology forward to the next limit.

What is the innovation

Humans can estimate the quality of different objects based on their appearance. We can perform this task, even with limited prior knowledge. In theory this makes humans the ultimate optical inspection tool, but we cannot concentrate on such a task for a long duration, nor do we deliver the speed needed for mass production. Computers, on the other hand, are well suited for the repeated tasks and can process large amounts of data at superhuman speed. However, teaching a computer to capture things the way humans do, and to let them tell apart features more complex and abstract than basic geometric shapes, has been the major challenge of machine vision for decades. Recent advances in Artificial Intelligence shifted the machine visions paradigm. Haiku Tech tapped into this development by designing a revolutionary system for optical inspection of ceramic or polymeric substrates for LED, fuel cell, electrolysis, and battery applications. Our proprietary Artificial Neural Networks enable real-time localization, recognition, and classification of functional defects on high-resolution visual data within a constrained machine cycle time.

Out of the lab. Into the market

Over the past two years, Haiku Tech has developed and tested several revisions of the optical inspection system, including numerous architectures of artificial neural networks. A standalone prototype machine has been put through two field tests at the production site of our project partner Elcogen AS, with a major hardware and software optimization step between the tests. Recently Haiku Tech has completed the design of an industrial standalone unit for 100% optical inspection of planar substrates up to 200x200 mm2 in size. Inspection can be performed at the resolution up to 3.5 micrometer/pixel. The system is offered as a standalone machine wich can be integrated in any production line. The first series of machines is scheduled for completion by Q4 2019.

Benefits of participation in the Horizon 2020 Programme

Thanks to the qSOFC project we were able to focus on finding the best hardware and software solutions for this technological challenge which remained unaddressed to date. Our initial market and technology research resulted in development of a new AI-based optical inspection system. Without the initial support, provided by the project members, and a fruitful cross-border collaboration with our partners, we might not have explored this new development as profound as we have done now. Participation in the qSOFC project and its results thus far comprise an important milestone for Haiku Tech as a company. It opens a whole new range of growth opportunities on the European market.

This innovation was funded via H2020 project qSOFC

Team behind the innovation

/futurium/en/file/haikuteampicturejpghaiku_team_picture.jpg