One of the significant causes of the undesirable stoppages and production defects is human error. It is estimated that people are responsible for 23% of such incidents. However, tests show that the number of production errors and pauses can be reduced by implementing AI in manufacturing powered vision systems in everyday operations.
AI for Proactive Maintenance in Wood Processing
For example, in the field of wood processing, AI can be used for proactive maintenance and result in a 30-50% reduction in downtime, lower maintenance costs, and increased overall equipment effectiveness. AI can also be effectively used for quality control purposes. Naturally, manufacturers are turning to AI technologies to combat the rising costs, as these technologies offer a competitive edge by reducing associated costs.
“The human factor accounts for a significant portion of all production errors. For instance, in one of our factories, we sorted wooden parts according to certain parameters. A human performs this task subjectively, adapting to the situation. On the other hand, AI performs everything consistently, without any deviations, reducing the risk of downtime”, says Augustas Urbonas, Head of Computer Vision Group at VMG Technics, a part of VMG Group, a global investment company currently operating 20 wood processing and furniture manufacturing companies in Europe.
How AI-Powered Vision Systems Improve Production Efficiency
Due to the nature of the work, financial losses are almost inevitable in the parts of production that humans traditionally handle. These repetitive tasks require a high degree of attentiveness, which people cannot maintain for an extended period. On the other hand, an AI can handle them perfectly.
According to Mr. Urbonas, the decision to implement AI in manufacturing tools was made because of their ability to perform tasks that require consistency and repetition. At Klaipėdos mediena, a woodworking plant part of VMG Group, they are responsible for segmentation and detection, as individual pieces must be separated according to specific indicators. AI-driven detection systems, combined with robotic vision technology, led to a 33% increase in productivity—from 16.3 to 21.76 square meters per hour. Standard packaging speed also improved, rising from 9 to 12 units per minute.
"We have been working on implementing these systems for about 2-3 years. The implementation itself is the easy part of the process. It takes longer to refine and adjust them, as numerous small details emerge that require consideration”, he says.
AI Anomaly Detection Reducing Production Stoppages
VMG Group estimates that production lines are typically un-operational for approximately 2–3 hours per week due to various unnoticed issues. At another of its companies, Sakuona, a curved plywood manufacturer, a deep learning model, the Siemens SIMATIC S7-1500 TM NPU, was used to solve this problem. It was trained using anomaly detection algorithms and data from previous years. In practice, this works by taking numerous photos of every item with dozens of cameras installed in the production line and feeding this information into the system.
“In this case, AI is tasked with anomaly detection and preventing production jams. This is achieved through specialised algorithms and by utilising existing operational data”, explains Mr. Urbonas.
Future AI Developments in Manufacturing Operations
His team is already planning to implement more AI in manufacturing powered tools. It is especially keen to use them to identify potential problems and avoid pre-emptively costly production line stoppages.
“I think we have learned a great deal throughout this process. The systems we use have only improved and become more complex. We envision a chatbot application enabling workers to communicate directly with machines, making the working process even more efficient.”
