Tekin Gulsen, CIO/ IT Director, Brisa Bridgestone Sabanci
Smart manufacturing, which is enabled and empowered by many new and exciting digital technologies, can have a real impact on a company’s manufacturing processes and its bottom line. Breakthrough improvements can be realized from quality inspections to workers’ safety, from logistics to maintenance.
One of the rapidly developing technologies that have many different application areas is image recognition and analysis. When coupled with machine learning, where an algorithm can learn and improve its understanding, image processing is more capable than traditional image processing techniques.
For such a system to be effective, initially, the system needs to be trained with the training data so that the image conditions can be classified. After a model is developed, it is tested with a test pool for its effectiveness and accuracy. However, if we were to use this system in production where many images are captured during the operations, it’s not feasible to transfer each video or image to the analytics system for processing. This is where the edge computing comes into play so that each captured image can be processed locally on a simple PC (mostly attached to the camera) with the model initially developed.
As a result of the analysis, the system should trigger an action to close the loop in the process.
Breakthrough improvements can be realized from quality inspections to workers’ safety, from logistics to maintenance
There are many application areas of such technologies, especially in quality inspections, safety and ergonomics, and warehouse operations.
In quality-related scenarios, machine-learning-based image analysis systems can be utilized to support or eventually replace a manual or visual control done annually or visually by a worker. It can lead the improved quality and productivity since an intelligent system replaces a repetitive and error-prone task. With the support of high-quality cameras, quality inspection can be done in more detail and more accurately.
Another exciting application area is facial recognition. There are many technologies in the market today that claim to determine your mood from an image. This facial analysis technology can be used to determine a worker’s level of focus and tiredness continuously. The system can generate a warning for the employee and the manager if the risk level is high to avoid any accidents of mis-productions.
Warehouse operations or any physical movement processes within the plant carry the risks of forklifts or trans-pallets, causing safety incidents. To avoid accidents in these operations, all the moving vehicles can be equipped with an image processing system that determines human beings around the car and distinguishes them from other machines or products in the environment. This smart image processing system can trigger an alert or even halt the vehicle in case of risk of danger.
The technology has matured so much that many solution partner companies claim that if you can perform a task with your eyes, this system can also deliver it at least as well as a human.
Like all other technologies, the critical success factor is to start by defining the problem or the business need so that the outcome of this technology can support a specific smart manufacturing initiative in your industry.