Traditional manual timber stack measurement is a slow, manual, labor-intensive process, plagued by inaccuracy and safety risks for workers in harsh weather. Dralle A/S, a leader in forestry digital technology, faced critical limitations with their legacy timber stack measurement system. But their initial system, an x86 platform, couldn't keep pace with modern AI demands, therefore the company partnered with Aetina.

The limitations were clear: slow machine learning execution, insufficient processing power for accurate log end detection, and a bulky size unsuitable for their camera box. These constraints ultimately rendered advanced AI and machine learning unusable on the existing platform, pushing Dralle towards a crucial turning point: the need for a solution capable of harnessing advanced AI and machine learning capabilities.

From x86 to Nvidia Jetson
To unlock the power of AI and achieve the desired accuracy and efficiency in a compact solution, Dralle partnered with Aetina. Their edge computing platform, featuring M.2 and GigE expansion ports for seamless camera integration, provided a robust hardware foundation specifically designed for the Jetson module. Additionally, Aetina’s expertise in Edge AI integration proved instrumental in enabling a smooth and quick migration (completed in just three months) from Dralle’s existing industrial x86 system to a Arm-based platform with Nvidia Jetson at its core. Leveraging Nvidia Jetson’s GPU, Dralle significantly enhanced its sScale system thanks to:

--- Increased Efficiency: Machine learning algorithms now detect up to 1000 unique log ends per frame in just 200ms, significantly reducing measurement time and manual intervention.
--- Enhanced Accuracy: The switch from classic computer vision to machine learning delivers highly precise log end detection with minimal false positives.
--- Improved Safety: Automated measurements eliminate the need for workers to leave the car, reducing safety risks.

Nvidia Jetson, with its low power consumption and unified memory architecture, played a vital role in this transformation. The compatibility of Jetson with standard AI/ML libraries, and its provision for expansion ports, prompted the decisive transition from the initial x86 platform. Aetina’s Jetson-powered edge device delivered the key advantages that drove the switch. Nvidia Jetson’s GPU and Deep Learning Accelerator (DLA) enabled Dralle to deploy sophisticated machine learning models for real-time log end detection and measurement, which represents a significant leap forward compared to the limitations of classic computer vision techniques used previously. Furthermore, Jetson Orin’s small form factor seamlessly integrated into Dralle’s existing camera system, maintaining a compact and efficient overall solution. Another point was that Jetson Orin’s CPU, GPU, DLA, and Programmable Vision Accelerator (PVA) provide Dralle with a scalable platform for future AI advancements. This ensures their system can adapt and leverage even more complex AI models as their needs evolve. Aetina facilitated the integration of Jetson with Dralle’s infrastructure, aligning seamlessly with Nvidia’s software ecosystem, which includes Metropolis for intelligent video analytics, TAO for AI model training, DeepStream for real-time streaming analytics, MMJ for multi-modal joint reasoning, and Isaac ROS for robotics applications. The software ecosystem offers pre-built tools and optimizations specifically designed for running AI applications on Nvidia hardware.

Summary
Aetina’s collaboration with Dralle exemplifies how they empower customers to achieve excellence in AI-driven industrial applications. Dralle’s successful migration to the Jetson platform not only revolutionized their timber measurement process but also positioned them as leaders in their industry’s digital transformation. With Aetina’s continuous support, Dralle is well-equipped to explore scalable solutions utilizing more complex AI models, paving the way for continued advancements in AI-powered forestry solutions.

Source: invision-news.de