Automation technology has been increasingly applied in the time of automation and AI. In the past decade, industrial robots have become an unstoppable trend on the production line. Many robots have integrated vision systems, using machine vision technology to detect, identify and measure objects to collect necessary information for subsequent robot movements.
As we all know, robots are superior to humans in dealing with repetitive work for hours, because they can rely on the set program. However, a disorderly environment can be extremely difficult for them. On the factory floor, the robotic arms were only able to grab objects in a fixed position. But with 3D vision, robots can perceive, analyze objects and the environment like we do. Objects placed at random places can be 3D scanned, and the depth information is obtained, along with advanced AI algorithm, the shape, volume and position of objects can be precisely detected.
In many automation applications, such as sorting, assembly and palletizing, the robot may need to complete a series of tasks such as target perception, motion planning, and picking planning. Take object grabbing as an example, the vision system needs to identify and locate the target object, provide the robot with the type, pose and other features of the object. So the accuracy of the captured information is critical for the grabbing. Below in the picture, Revopoint’s Rayscan 3D camera is grabbing steel workpieces of different sizes and shapes and will put them into the basket. As the 3D camera captures accuracy up to 5μm, it can detect around a million 3D points of the workpiece and create a 3D model, then the features of the workpiece will be analyzed by AI algorithm to help it grab up the metal piece.
In many other cases, 3D vision is able to help improve production efficiency and quality, and cut costs. During the online shopping carnivals like China’s Singles Day or the incoming Black Friday, workers have to sort out of tens of thousands of packing boxes in the warehouse, with 3D vision, three-dimensional data of the packages can be obtained and analyzed in a very short time, packages will be automatically sorted out and go to the right place.
