Globally, agricultural fields waste a massive amount of pesticides each year, and NVIDIA’s “Edge AI” is solving this problem! It equips sprayers and robots with “eyes” that use cameras and sensors to accurately identify weeds, applying pesticides only where necessary, or even directly pulling them out. The underlying Jetson module is powerful and durable, designed to withstand the dusty and wet conditions of agricultural environments.
Edge AI: NVIDIA-based Embedded Systems for Agricultural Technology
Edge AI: NVIDIA based Embedded Systems for Agricultural Technology
Source:NVIDIA News
Globally, 3 million tons of pesticides are sprayed on agricultural fields each year, yet only a small fraction of these chemicals is actually necessary. Now is the time to take action! Many suppliers of sprayers, tractors, and agricultural robots have recognized this issue and are taking various approaches to address it. For example, some sprayers are equipped with cameras and sensors that capture ground images and identify weeds, applying pesticides only in necessary areas; another solution, which operates on a similar principle, completely avoids the use of pesticides—agricultural robots pull out the identified weeds. Both methods help reduce pesticide usage while increasing crop yields.
This example illustrates the immense potential of machines equipped with “vision” and “understanding”—for instance, in ensuring global food supply. But the question is, how can machines learn to identify weeds?
Pre-trained AI Algorithms Empowering Edge Intelligence
Edge intelligence is a hot technology today. With the help of AI algorithms, data can be processed directly at the point of generation (i.e., near the sensors). Commonly used sensors include 2D and 3D stereo cameras, LiDAR, and radar. The generated data is processed through pre-trained neural networks during the inference process. The so-called “inference process” is essentially the process by which software autonomously draws conclusions from the collected data: sensor data is analyzed and evaluated at the edge, while the neural network continues to learn from new data—such as identifying new components in the production process.
Sensors, Software, and AI-Enabled Embedded Computers
Implementing edge intelligence requires three core components: software, sensors, and AI-enabled embedded systems. Among these, the embedded system serves as the hardware core, capable of processing data in real-time and making intelligent decisions based on it. Today, chip manufacturer NVIDIA’s technology is often used for such inference tasks. The NVIDIA Jetson product line offers system-on-modules (SoMs) with varying performance levels, integrating CPU and GPU technologies. With a parallel processing architecture, these SoMs can quickly and efficiently run software for autonomous machines and devices, especially capable of processing data from multiple high-resolution sensors with virtually no latency.
Another major reason for choosing the NVIDIA Jetson platform is its development kits. Companies can leverage these kits to kickstart software development and seamlessly complete the development process on production equipment. The platform also provides a wealth of libraries and specific application frameworks, effectively reducing development costs. Additionally, this series of SoMs is compatible with the Robot Operating System 2 (ROS2) middleware—ROS2 has become an ideal tool for computer vision applications, used to control and coordinate numerous nodes. This middleware features a modular structure, capable of not only processing and evaluating sensor data but also controlling actuators.
NVIDIA’s latest system-on-module is the Jetson AGX Orin, which boasts an AI computing power of 275 TOPS (trillions of operations per second), 64GB of memory, and adjustable power consumption ranging from 15 to 60 watts. The powerful CPU combined with the NVIDIA Ampere architecture GPU creates a unique combination, enabling new computer vision applications across various industries, with typical applications including hazard detection, environmental perception, intelligent video analysis, and autonomous system control.
Embedded Computers Can Be Dust and Water Resistant
To ensure that NVIDIA’s AI technology operates normally in harsh environments (such as those encountered by production equipment or agricultural robots), specially designed hardware is required. A few global suppliers have taken on the task of “adapting NVIDIA technology to harsh environments,” launching embedded computers based on NVIDIA Jetson with an IP67 protection rating. For example, Syslogic combines system-on-modules (SoMs) with self-developed carrier boards, high-durability enclosures, and sophisticated connector technology; their rugged computers utilize a passive cooling design, suitable for a wider temperature range. The company also collaborates with sensor manufacturers that have similar durability and reliability requirements.
Thus, even in harsh industrial application scenarios, the potential of AI can be fully unleashed. By using Power over Ethernet (PoE) or Gigabit Multimedia Serial Link (GMSL) interfaces, sensors such as LiDAR, radar, and cameras can be connected to embedded systems, enabling complex computer vision applications—opening new opportunities for global enterprises.
Source: Bilingual Smart Agriculture LearnSomeAgtech
