From the Blog

: Now more than ever, technology buzzwords such as AI or artificial intelligence, machine learning, deep learning, Internet of Things (IoT) and Industry 4.0, are drawing the interest of several industries for specific applications, or to solve specific problems.

Recently Pleora Technologies exhibited at CRAV.ai (Collaborative Robots, Advanced Vision & AI Show) and instantly acknowledged that the conference attendees included a variety of participants in a vast range of industries such as agriculture, manufacturing, and animal welfare – all looking to solve problems specific to their line of business, through quality inspection systems using machine learning and AI. Knowing there is an abundance of AI technologies, each aiming to solve complex technical problems, excitement thrusts forward the interest of many manufacturers looking to improve their machine vision systems and move into the next phase of robotics.

 

However the challenge many face is twofold; there is a general lack of awareness in how AI systems interface with existing equipment in a quality inspection setting and a misconception that the customer will be responsible for integrating the AI algorithms. Market data highlights the importance of these AI inspection applications, yet many companies still need to define their approach in how to augment inspection systems with AI and sensor networking while working towards their projected needs for Industry 4.0. In addition, system integrators have been focused on lowering operational costs, addressing bandwidth requirements and designing with scalability in mind. Now, they will also be required to face additional challenges with no real ‘standard’ for how best to deploy AI and algorithm training.

 

The common question remains, what is the best solution to deploy AI within existing machine vision infrastructures? By evolving to a high definition inspection system, manufacturers towards networked embedded architecture, where smart devices make a decision and send data on to other devices and local or cloud-based processing. Local decision-making will significantly reduce the amount of data to be transmitted back to a central processor, therefore allowing the system to share real-time, high-bandwidth data, while developing a more efficient, IoT ready, scalable system. This processing technique allows for a simplified method to integrating the basics of artificial intelligence within an existing traditional inspection system while developing networking protocols for device and sensor communication, via GigE vision and OPCUA.