Articles
“There are challenges that come with implementing AI and one of the key ones is a lack of understanding of AI and its potential benefits as well as finding the right starting point. So how do enterprises find the right use case and define the strategy?” asked Dr Nandini Chakravorti, chief engineer in the digital engineering group at the Manufacturing Technology Centre (MTC) in Coventry.
Organisations like the MTC have been working on different AI solutions and use cases for a number of years and so realise the vast benefits that can be reaped from implementing them. They have also discovered that the key to AI and, indeed, Industry 4.0 in general is data.
Chakravorti said: “AI and data analytics are key enablers to generate value from data and there are a number of ways in which AI can improve manufacturing. It can help companies detect new consumption patterns and deliver highly personalised products, it can transform how inspections are conducted by providing more eyes and ears, it can be used to address the growing challenge of product defects and recalls, and AI algorithms can be used to optimise supply chains, helping companies to anticipate market demands.”
Although the benefits of AI are clear there is still the question of where companies should start with its implementation. Chakravorti said: “Our suggestion is always to start small or simple and think for big. For example, start with monitoring your production then go a little bit further into why that happened and when will it happen. You also must have a strategy for implementing AI and don’t forget that the most challenging aspect is the change management.”
Low-cost solution
One of the MTC’s most recent solutions is the Flexible Automation Through Machine Learning demonstrator. This system, which was built using a combination of relatively low-cost camera hardware with a collaborative robot platform, combines machine learning and robotics to enable the robot to successfully grasp components and assemble the example product.
Most automation technology in industry is specifically programmed for a given task and cannot accept any input variation. Changing the process even slightly can require major investment and so prevents manufacturers from being flexible. The aim with the demonstrator was to create a low-cost assembly system that removes the need for expensive fixed tooling.
Tom Winter, senior research engineer at the MTC, explained: “This cell really demonstrates a flexible solution. The AI system is able to, on request, process an image from the intake tray and individually identify the components within that tray. By integrating multiple deep-learning models together the system not only provides a component’s location but also validates the rotation of the component within the tray.
“We are able to transform the robotics cell much quicker than we would in a traditional system as the robot is able to take the part from within an open input tray rather than having specialised fixtures that are built to hold the components in the correct rotation. The same tray can be reused with minimal effort both from the AI side of needing to retrain and reupload a new network neural model and then from the robotics side of being able to teach the robot how to collect that component.”
For more content like this sign up to our free webinar series – Engineering Futures.
Content published by Professional Engineering does not necessarily represent the views of the Institution of Mechanical Engineers.