Machine vision is a rapidly developing branch of artificial intelligence. Simply put, machine vision uses machines to replace human eyes for measurement and judgment. The rapid development of the internet has fueled the explosive growth of the logistics industry. Not only do annual shopping festivals like JD.com and Taobao put immense pressure on logistics personnel, but even daily food delivery and supermarket deliveries are seeing a surge in manpower. Meituan Waimai is deploying unmanned logistics, and JD.com has begun using robotic picking. Robots are bringing immense convenience to people's lives, and they are gradually becoming market darlings.
Today, robots of all types are ubiquitous around us, playing a vital role in manufacturing, transportation, and daily life. Examples include robotic mobility scooters and robotic vacuum cleaners. The technology that gives these robots "intelligent" eyes is machine vision.
Machine vision is a rapidly developing branch of artificial intelligence. Simply put, machine vision uses machines to replace human eyes for measurement and judgment. It is based on biomimicry, for example, simulating the eye by acquiring images through visual sensors, followed by image processing and recognition by an image processing system.
Classification of machine vision
Machine vision is mainly divided into three categories:
Monocular vision technology, which involves using a single camera to acquire images, typically only produces two-dimensional images. Monocular vision is widely used in the field of intelligent robotics. However, due to limitations in image accuracy and data stability, it needs to work in conjunction with other types of sensors, such as ultrasound and infrared sensors.
Binocular vision technology is a method that simulates how human eyes process environmental information. It uses two cameras to capture one or more images from different perspectives to establish the three-dimensional coordinates of the object being measured. Binocular vision technology can be broadly categorized into areas such as visual control of robotic arms, visual control of mobile robots, and visual control of unmanned aerial vehicles (UAVs) and unmanned surface vessels (USVs).
Multi-view vision technology refers to the use of multiple cameras to reduce blind spots and lower the probability of false detections. This technology is mainly used for measuring the motion of objects. In terms of hand-eye coordination in robotic arms, multi-view vision technology can overcome blind spots in object capture, making the robotic arm's grasping more effective. In the field of industrial robot assembly, multi-view vision can also accurately identify and locate the object being measured, thereby improving the intelligence and positioning accuracy of assembly robots.
