This page contains a Flash digital edition of a book.
ARTIFICIAL INTELLIGENCE


NDT Read more on NDT at https://goo.gl/79WszT or scan the QR code


behaviours, such as eating and drinking or programming the GPS navigator while driving, were accurately classified using a novel method for converting accelerometer time-series data into image representations, which were then fed to a deep neural network.


Artificial intelligence revolutionises NDT


F


ujitsu Laboratories of Europe has developed an innovative AI technology that substantially accelerates the Non-Destructive Testing


(NDT) inspection procedures in complex manufacturing processes. The technology applies cutting edge deep learning, image and signal processing techniques to analyse NDT data and identify patterns that may indicate manufacturing defects. It is relevant anywhere non-destructive inspection techniques such as ultrasound, X-ray or imaging are used, including electronics, the automotive industry, aircraft production, pharmaceuticals, transport and infrastructure. The new approach transforms current


Quality Control (QC) procedures, automatically identifying interesting scan areas for defect inspection and typically reducing manual inspection procedures by up to 80%. Additionally, it continues to learn after deployment, improving performance on an ongoing basis. The deep learning exploits neural


networks to process large amounts of image data to detect relevant patterns in NDT ultrasound scan data. Specialist manual inspection can therefore be rapidly targeted to explore potential defects requiring the attention of an expert technician. As a result, potential bottlenecks in the production process are removed, with the potential to scale


6 /// Electronics Testing 2018


up production and make significant efficiency improvements. “We developed a generic machine learning engine for pattern detection, using a process that translates any raw data analysis problem into one involving image pattern recognition,” says Dr Adel Rouz, Executive Vice President of Fujitsu Laboratories of Europe. “Working with manufacturers, we can rapidly tune into a specific application, thanks to its ability to learn from just a few training examples. This significantly minimises the amount of annotated data needed.“ In one application, it was applied to improve the retrieval of 3D CAD models from massive databases, helping to accelerate product design and enhance QC. In another example, the technology was applied to a social innovation application, detecting driver behaviour via a discrete wrist-worn acceleration sensor. Potentially dangerous


AI ACCELERATORS FOR SMARTER CARS BrainChip Holdings, a developer of software and hardware accelerators for advanced artificial intelligence (AI) and machine learning applications, has shipped its first BrainChip Accelerator card to a major European car manufacturer. The BrainChip Accelerator is a branch


of artificial intelligence that simulates neuron functions. It will be evaluated for use in Advanced Driver Assist Systems (ADAS) and Autonomous Vehicle (AV) applications. The BrainChip Accelerator increases


the performance of object recognition provided by BrainChip Studio software and algorithms. The low-power accelerator card can detect, extract and track objects using a proprietary technology called a spiking neural network (SNN). It provides a seven- fold improvement in images per second per watt, compared to traditional convolutional neural networks accelerated by Graphics Processing Units (GPUs). “This is an exciting first evaluation


of BrainChip Accelerator that was released just last month,” commented Bob Beachler, BrainChip’s Senior Vice President for Marketing and Business Development. “Our spiking neural network provides instantaneous “one- shot” learning, is fast at detecting, extracting and tracking objects, and is very low-power. These are critical attributes for car manufacturers processing the large amounts of video required for ADAS and AV applications.” EE


❱ ❱ Fujitsu uses AI to point the operator to the regions of the scan that need to be inspected


Page 1  |  Page 2  |  Page 3  |  Page 4  |  Page 5  |  Page 6  |  Page 7  |  Page 8  |  Page 9  |  Page 10  |  Page 11  |  Page 12  |  Page 13  |  Page 14  |  Page 15  |  Page 16  |  Page 17  |  Page 18  |  Page 19  |  Page 20  |  Page 21  |  Page 22  |  Page 23  |  Page 24