Machine learning enhances manufacturing productivity

Introducing artificial intelligence to your manufacturing is not about robotic workers, but the introduction of machine learning to improve performance. Major advances in situational awareness developed from AI used in autonomous vehicles will help manufacturers improve productivity. Early applications have included augmented reality, cobots and predictive maintenance systems.

Key to this is machine learning (ML), using layered algorithms to process data and learn in the same way as humans. These are still early days for AI adoption, and it will not replace rule-based programming for many applications. However, according to McKinsey, introducing digital predictive maintenance, can reduce maintenance costs between 10 and 25%. Furthermore, asset availability is also likely to improve by 5 to 15 percent. Other suitable applications should expect to deliver similar productivity improvements. They also release skilled employees to undertake higher level work.

AI allows deskilling of sections of the workforce to help organisations cope with smaller, less trained workers for some tasks. However, it is important to recognise that AI black boxes can also make sub-optimal decisions. Moreover, management must keep the core skills needed to run their businesses and cannot abdicate ultimate control or responsibility.

Smart visualisation

Using connected technology, operators can interact with machines and learn how to use them more quickly. Mitsubishi Electric has developed a maintenance support technology using smart glasses as user interfaces for machine monitoring. Operators can view up-to-date machine and production information, such as progress, remaining time, machine status or overall equipment effectiveness (OEE). Remote colleagues can also send 3D images and information to the operator in the event of problems. Visualisation technology supports predictive maintenance but can also aid with training for operators or maintenance staff.


Human operators can easily distinguish between different objects in a container and pick the correct one. For a robot, this presents a challenge for vision technology to accomplish. Many conventional robot applications require complex and expensive feeder systems to correctly orientate components.

AI supports the implementation of cobots (collaborative robotics) in the workplace using enhanced machine vision systems. Smart-control AI technology enables industrial robots to rapidly grasp and adapt in real time to changing conditions of target objects. It simplifies automation tasks, even during changing conditions, such as adapting to the changing shape of a non-rigid objects.

They use machine learning algorithms to optimise the performance of the robot to give them ‘self-awareness that supports interactions with others in its environment. The automatic generation of optimal control algorithms through deep reinforcement-learning frees designers from having to redesign complex control algorithms.

Predictive maintenance

For many plants, the calendar is the basis for machine maintenance. However different machines and even different parts of the same machine may require maintenance at different times. Component failure can also lead to unplanned downtime, lost production and waste.

Predictive maintenance uses smart sensors to monitor and evaluate machine parameters, like temperature, vibration and running time etc. Data analysis with the help of cognitive AI algorithms helps early identification of likely causes of failure. Smarter HMIs will also ensure that operators see this critical information on machine condition, allowing timely preventative maintenance.

Using edge computing and cloud computing, machine learning will start to positvely impact on areas like quality control, and supply chains and flexible manufacturing.