Adapting to technological change is essential for leaders in the high-stakes arena of industrial manufacturing and non-destructive testing. Our client, a leader in the field of non-destructive testing machinery manufacturing, recognized the potential of artificial intelligence to revolutionize their product. As industry innovators, they strive to transform common practices in NDT into more efficient processes enhanced with AI.
While working with this client as a sparring partner for their in-house AI research for several years, we at Neuroforge were always inspired by their dedication and commitment to move the industry forward by constantly innovating their solutions. These are the common values we share, which form the basis of our cooperation.
When the time came to turn their findings into a fully functional AI component to enhance their NDT machinery with visual surface inspection, we were happy to assist them as their development partner.
Brief Summary:
In their quest to remain a market leader in non-destructive testing machinery development, our client sought to enhance their capabilities through artificial intelligence. After developing a concept for an AI extension for their NDT machines, this partner worked with Neuroforge to leverage our expertise in fully developing and implementing a robust AI solution integrated into the machine's infrastructure.
Teaming up with Neuroforge, the client embarked on a transformative project that began in June 2020. Starting as a sparring partner for in-house AI research, the collaboration evolved into the development of a fully functional AI-driven application integrated into their existing systems.
The partnership yielded software tailored specifically for the client’s machinery, significantly enhancing its capabilities, improving the precision of material classification, and strengthening its position as a market leader.
Vision
Recognizing the need to advance beyond traditional methods, our partner had previously explored integrating AI into their systems. They developed an initial machine learning model and user interface concept aimed at automating the classification of ceramic balls for quality assurance — a pivotal step in their testing capabilities that promised to significantly boost both testing accuracy and efficiency, setting a new standard in testing machinery.
This initial effort yielded a prototype, and the client quickly realized the need for a sophisticated, fully operational AI solution to maintain their competitive advantage. While their existing systems were already effective, they envisioned integrating advanced AI capabilities to enhance accuracy, minimize errors, and reduce operational costs for their customers. This is when the partnership with Neuroforge evolved from R&D and sparring partner activities to developing a fully operational AI component solution.
Together with us as their development partner for this project, our client set out to create a new generation of AI-enhanced testing machinery, aiming to improve on their existing state of the solutions to build the next generation of testing equipment. This collaboration was driven by their visionary approach and the necessity for specialized expertise in AI development and system integration. We brought the necessary technical prowess to expand the client’s AI concepts into robust components capable of sophisticated, real-time material classification.
Solution
Our collaboration has brought to life a customized software solution, tailored to meet the needs of our client and their customers. This solution integrates an advanced AI component within their new generation machines, significantly enhancing the system's analytical capabilities and creating a new experience in using NDT machinery.
The basis of the AI component was underpinned by implementing a RESTful API framework. This interface facilitates a seamless integration of the AI component with the existing machine systems. The primary advantage of utilizing a RESTful API is its capacity to offer a flexible, web-based interface that enables various software modules to communicate efficiently and securely across the machine's network. The RESTful API acts as critical middleware, effectively bridging disparate systems and ensuring that they can exchange data and commands without compatibility issues, akin to an interpreter who ensures fluent communication between speakers of different languages.
The AI component was collaboratively developed to enhance the capabilities of the client’s existing laser-based computer vision technology. This system was already proficient at generating images of ceramic balls for defect inspection. Together, we leveraged this scanning system by integrating an AI that analyzes each image, identifies defects, and classifies the balls into categories such as 'OK' or 'not OK.' This enhancement utilizes a deep learning model that mimics human pattern recognition, significantly augmenting the existing technology's capability to pinpoint and classify defects more effectively.
To address the need for greater transparency in the testing process, the AI system was designed with an explainability feature that provides operators with clear insights. When a ball is labeled as 'OK' or 'not OK,' the AI highlights specific areas on the ball’s surface, indicating detected defects. This not only shows what was identified but also where it is located, which is crucial for our client’s customers who are looking to pinpoint manufacturing flaws. This level of detail allows them to make targeted improvements in their production line, enhancing the overall quality assurance process.
In the development of the user interface for the AI component, we closely followed the design specifications provided by the client while also contributing our design expertise to enhance the system’s functionality. The interface showcases a 3D representation of the ceramic balls, with an AI explainer that highlights defects directly on the image. Below this visual display, a historical log of test results is available, allowing operators to easily track trends and recurring issues. This not only improves the user experience but also boosts operational efficiency by making data more accessible and actionable.
The software's adaptability is a key feature, designed to classify a broad range of raw materials. This flexibility is realized through a feature that facilitates the easy import and integration of new AI models, thereby eliminating the need for extensive reprogramming. Should our client need to inspect a new type of item, they can effortlessly integrate a specifically tailored AI model.
In collaboration with the client’s development team, which oversees the machine's control systems, we seamlessly integrated the AI application into the overall system. This integration involves automating the data flows between the AI component and the machine's main computer system via PLC APIs, enabling the automatic processing and utilization of classification results. We ensured that the
AI systems could deliver real-time classification results that fit performance requirements. For this, we used various performance profiling and optimization techniques to achieve high throughput in the AI pipeline running in the final application.
To ensure comprehensive, high-quality training for the AI models and optimize the necessary grid search process for finding optimal hyperparameters, we employed tailored MLOps processes that focused on the orchestration of research, quick AI model development, training, and enhancing AI explainability.
Outcome
The partnership with Neuroforge delivered a new generation of AI-enhanced non-destructive testing machinery that is making a significant mark in the industry. This advanced technology has not only met but exceeded expectations, offering exceptional accuracy and efficiency. The client’s customers in various industries are already reaping the benefits. These innovations have led to successful deals and endorsements from industry leaders, proving the real value and competitive edge these AI-enhanced machines bring to the market.
As Neuroforge, we are excited and proud of this collaboration, which highlights the client’s commitment to innovation and Neuroforge's expertise in AI development. Together, we have created a product that not only enhances the client’s offerings but also sets new standards for quality and efficiency in industrial testing.
This project has been a transformative journey, and the success achieved is a testament to what can be accomplished when vision and expertise come together. The AI-enhanced machinery from the client is now a game-changer for industries looking to enhance their quality assurance processes.
If you are looking to bring an exceptional AI and BigData expertise to your service or product, we at Neuroforge have you covered. We will help you in uncovering possibilities AI can bring to your business and then we empower you to turn them into real, custom-tailored solutions.