Identification of Ground Engaging Tools (GET) Using Computer Vision

Abstract

Within the mining sector, Ground Engaging Tools (GET) are developed to operate under severe conditions of abrasion and impact. These tools are made of extremely rigid materials to withstand such harsh operating conditions. However, when a breakage or loss of an GET occurs within the mining process, it becomes a problem due to the negative impact it has on the subsequent stages of the process. For instance, it may cause the crusher to jam, resulting in productivity losses and potentially critical safety hazards when trying to remove this type of material from the crusher. In this context, this study proposes the use of artificial intelligence integrated with strategically placed cameras to continuously monitor and evaluate the integrity of these GET during the production process, specifically throughout the ore loading operations performed by the Letourneau equipment. The proposed solution employs convolutional neural networks (CNNs) to analyze real-time images, observed by an external camera positioned on the side of the machinery, enabling precise identification of each GET on the equipment’s shovel. This approach allows for immediate detection of any damage or loss, triggering alerts to prevent potential future issues. The obtained results indicate the feasibility of using this type of computer vision strategy to mitigate or prevent various problems in the ore extraction process.