Musashi AI consortium aims to quickly apply AI to Industry 4.0

Musashi AI consortium created to apply AI to Industry 4.0

Israeli technology pioneer Ran Poliakine and Musashi President and CEO Hiroshi Otsuka announce the formation of Musashi AI to bring Industry 4.0 into reality in months versus the expected years. (Credit: Business Wire)

TOKYO — Although there has been a lot of talk about Industry 4.0 and advancing manufacturing with automation and artificial intelligence, many projects have focused on only one technology or another. Yesterday, Musashi Seimitsu Corp. announced the formation of the Musashi AI consortium at the AI Expo Tokyo to quickly develop products for smart manufacturing.

“The benefits that artificial intelligence brings to the manufacturing industry are clear, but it takes collaboration, commitment and focus to make the promise of the Industry 4.0 revolution a reality,” stated the company. “The ability for machines to cooperate with humans in real time sparks innovation and creates new efficiency and agility in manufacturing processes, leading to reduced costs, larger revenues, and improved customer experiences.”

International partnership

Musashi Seimitsu Industry Co. is a global Tier 1 supplier of parts for automobiles and motorcycles. The Toyohashi, Japan-based company designs, develops, and makes powertrain products and is an affiliate of Honda Motor Corp.

At AI Expo Tokyo, Musashi Seimitsu said it is creating the Musashi AI consortium with Poliakine Innovation and SixEye Interactive. Poliakine Innovation founder Ran Poliakine has founded several technology companies in Israel, and SixEye Interactive Ltd. is an AI software company in Israel. The partnership between the Israeli and Japanese companies follows an intergovernmental agreement signed in January.

“As we enter this new world of disruptive innovation, organizations that embrace Industry 4.0 will enjoy the rewards of greater productivity, rapid innovation, reduced costs and empowered employees,” said Hiroshi Otsuka, president and CEO of Musashi Seimitsu. “We created Musashi AI consortium to act nimbly and efficiently and to serve as a leader of this revolution. We are excited to build a vibrant new ecosystem around Industry 4.0 because this revolution requires strong partnerships amongst likeminded industry leaders.”

Monozukuri attitude

In Japanese, the term monozukuri refers to manufacturing, as well as the combination of technical know-how and a can-do spirit. The Musashi AI consortium intended to be a similar collection of capabilities.

“I’ve had experience in using AI in other startups, such as in medical robotics,” Poliakine told The Robot Report. “The CEO of Musashi and I sat down and discussed Industry 4.0, and we figured out that many of the challenges that I had gone through in terms of algorithms and sophisticated optics in the medical space were applicable to Industry 4.0.”

“Everybody’s talking about edge computing, especially in areas where processing is local for optics and security,” he said. “Many failures of industry are because of the challenges with AI and optics, but when we brought together our domain experience, we had positive results.”

The new consortium demonstrated two AI-powered prototypes at AI Expo Tokyo that it said took only months to develop rather than years.

“We took big buzzwords and made them a reality, and we’re showing working systems here in Tokyo,” Poliakine said. “We’re very proud of these initiatives.”

Cutting-edge visual inspection

The Automatic Inspection System (AIS) uses AI and optics for parts inspection and quality assurance (QA).

“Eight million people work in the automotive industry in Japan, and 20% of that workforce is devoted to QA,” said Poliakine. “It’s very labor-intensive, and the industry has zero tolerance for mistakes.”

“If you look at the operators for this core industry function, they take an object in hand — which could be ceramic, metal, etc. — look at it, and take maybe two seconds to decide if the part is OK,” he said. “Humans use their hands, eyes, and brain to make a decision of good, bad, or maybe, but they could be doing more productive, satisfying work.”

“Many of the approaches so far focus on the AI — if only you have enough images, you can statistically reach the right decision,” Poliakine said. “We figured out that we weren’t paying enough attention to the process between the eyes and the brain. We created a proprietary bionic eye to see the right wavelengths in the right conditions, adjusted to look at specific materials for defects.”

“It’s similar to the QA approach for semiconductor wafers. By applying it to specific tasks, they’re immediately an order of magnitude better,” Poliakine said. “The second part is using an algorithm using good data. Instead of 100,000 images on a very big GPU, we can actually look at something with CPU power. We’ve done it with Musashi on the production line. When you narrow down the solution, you gain a lot, which is very different from what others have done.”

“A robot feeds parts into a chamber that includes a GPU, some optics, and logic that helps to make a decision if a part is good or not,” explained Poliakine. “We’re offering it to customers that need quality assurance from one end, such as automotive manufacturers, to the other end, Type A suppliers. Parts from 0.5 in. to 5 in. in diameter can fit in the chamber. In can actually best the best QA workers.”

Robotics Summit & Expo 2019 logoKeynotes | Speakers | Exhibitors | Register

Musashi AI forklift

The Fully Self-Driving Automated Forklift (FAF) is designed to navigate autonomously and to improve warehouse worker safety by performing tasks that previously required human involvement.

“We’ve seen so many solutions claiming to do affordable self-driving vehicles, but our connection to the industry found that the reality was far different,” Poliakine recalled. “Eighty percent of the production facilities in the world aren’t clean rooms that can work for AGVs [automated guided vehicles]. There are people wandering around and a lot of dynamic changes.”

He claimed that fully autonomous vehicles are still in development, are typically too expensive for factories, and are more sophisticated than necessary. The FAF works with a central “brain” controlling multiple vehicles.

“The central brain can help with orientation, and the local car just deals with avoiding collisions and the last few inches,” said Poliakine. “Given that we’re working in a controlled environment, that gives us a lot of certainty about the total number of locations.”

“Ideally, the system that we’re developing will be a simple API [application programming interface] that can sit on any forklift. It’s more integration than construction of the technology itself,” he said. “Users really want to track how things are done, so we need to connect to ERP [enterprise resource planning] and other systems. It’s a simple form of the Internet of Things.”

Musashi AI plans to test and demonstrate FAF in a production environment later this year.

Bringing new tech to manufacturing

The Musashi AI consortium is open to additional partners and emerging technologies, but it first wants to focus on a vertically integrated value proposition, Poliakine said.

“With IoT and 5G, we’re going to see democratization of sensors — sensors everywhere,” said Poliakine. “We’re trying to get instant benefits in terms of data and managing multiple tasks.”

“I believe that what we’re doing is practical. We’re building actual machines that are affordable and do the work required,” he said. “We’re taking a combination of edge computing, machine learning, and optics — that’s the difference between just a theory and disrupting industry. We’ll start with industries that have little tolerance for low quality, such as high-end automotive, aerospace, some agriculture, pharmaceuticals.”

“What we bring to the market is Israeli high-tech experience with Japanese blue-collar tech for very practical solutions,” he said. “This is a big opportunity to work with mainstream industry and in a high-tech revolution.”

The post Musashi AI consortium aims to quickly apply AI to Industry 4.0 appeared first on The Robot Report.