Mitsubishi Electric participated as a global sponsor at ARC Advisory Group’s recent India Forum in Bangalore, which attracted over 250 delegates. In his presentation, Sunil Mehta, General Manager, Automotive Business Development Department, Factory Automation & Industrial Division, Mitsubishi Electric India spoke about Artificial Intelligence (AI) and how the technology can be used effectively in the manufacturing sector. As with many of the other presentations, this aligned well with the theme of this year’s Forum, Digitizing and Securing Industry, Infrastructure, and Cities.
According to the company, AI is about making machines smarter to improve safety, efficiency, and comfort. AI research includes advances in computer vision, speech, and audio processing, as well as data analytics. Key research themes cover the areas of deep learning and AI for object and action detection; and simulation of physical systems to enhance machine learning systems. These solutions are applied to a broad range of automotive and robotics applications, as well as building and home systems.
The key takeaways from Mr. Mehta’s presentation:
- The line between man and machine is blurring
- Artificial Intelligence uses specific skill sets; combines teaching+experience; works at different levels; and will increasingly be utilized in smart devices
- AI will drive the future of manufacturing
Understanding and Deploying Artificial Intelligence
At previous ARC India forums, Mitsubishi Electric spoke about the company’s e-F@ctory smart manufacturing solution and condition monitoring systems. Mr. Mehta discussed the basics of AI and deploying AI technologies; substantiated by case studies. He explained that, like a baby, AI must be guided and molded according to how we want it to perform. AI does what is taught, uses past experiences to predict an outcome, and only works within its skillset. Explaining its ubiquity, he said that AI is already being used in commercial airplanes (auto-pilot mode), and also by companies like Google in internet searches, PID loops and vision recognition. AI has specific skills in different areas, but these are not interactive.
Artificial Intelligence in the Manufacturing Context
Mitsubishi Electric is investing in AI research as the company believes this will help and support the manufacturing industry. AI can be used to adjust parameters, perform voice/face recognition, search for problems, and assist in both preventive and predictive maintenance. In a developing country like India where the wages are rising, automation and robotics play increasingly important roles. Mr. Mehta explained that AI-based manufacturing will exist at all levels starting from the machine level, where reaction time, server/network costs, safety/security are of paramount importance. Industrie 4.0, e-F@ctory, etc. will coexist, but AI will have a much larger role in the manufacturing context, said Mr. Mehta. Mitsubishi Electric has introduced Maisart (Mitsubishi Electric’s AI State-of-the-ART) – a compact AI system that offers scalability and can be adopted in manufacturing industries.
Mitsubishi Electric has already implemented its AI solution-on-a-chip in their air conditioning systems and now have plans to introduce speech recognition function into motion control and other factory automation products. He discussed successful case studies on driver and car health monitoring; how AI separates speech from background noises; preventive/predictive maintenance; and maintenance with augmented reality.
Synopsis of the Case Studies
Merging People and Machines
The dividing line between operators and machines is becoming narrower, he said. The driver monitoring system is symbiotic – it provides real-time information about the vehicle and the driver. This is very useful because if the driver is sleepy, distracted etc. the vehicle can take corrective action; or if a problem is detected in the vehicle, the driver is alerted.
Demonstrating how AI separates speech from the background noises, Mr. Mehta showed the video of a cocktail party. The company has developed algorithms to achieve 90 percent accuracy in separating individual speech. The same system can be used in a manufacturing environment where the vibrations and noise of the machinery often drown out everything else. Maisart’s proprietary declustering technology generates deep network encoding of the mixed speech and identifies the elements of each speaker and separates the sound files. “When it comes to people and machines, AI has to be there,” he said.
Automatic Picking in Logistics Centers
Automatic generation of optimal control algorithms through deep reinforcement-learning makes the life of today’s designers easy by avoiding redesign of complex control algorithms. For example, AI can help robots in logistics centers differentiate between different-shaped objects or adapt to flexible objects that change shape.
Speaking about how AI is used in preventive and predictive maintenance,
Mr. Mehta cited the example of the IoT-based sensor that fits on an induction motor for monitoring vibration, the current and temperature of winding that Mitsubishi Electric Germany developed in conjunction with the Schaeffler Group. The sensor monitors the motor and helps predict failures.
In various applications, AI can be utilized to:
- Automatically suggest appropriate maintenance and necessary tools
- Support links with purchasing systems to order needed parts
- Support voice recognition and augmented reality (AR) for hands-free interaction
Mitsubishi Electric has been a regular participant in ARC Industry Forums globally. At this year’s ARC India Forum, the company took the opportunity to share knowledge and best practices about AI and how it believes the technology will drive the future of manufacturing. The company also provided case studies to substantiate this. There is much more to come – fast force-feedback algorithms, automatic tuning of 3D vision sensor algorithms, and real-time diagnostics (wave form recognition).
With its technology expertise, continuous research into cutting-edge areas such as AI, and focus on innovation, Mitsubishi Electric appears to be well-positioned to help industrial organizations achieve operational excellence.