Testing more than 60 types of screws at the same time, Mingyue Lake Enterprise provides "eagle eyes and golden eyes" for the UHV industry
2024-08-11
한어Русский языкEnglishFrançaisIndonesianSanskrit日本語DeutschPortuguêsΕλληνικάespañolItalianoSuomalainenLatina
What kind of "eagle eyes" does it take to command the manipulator to accurately grab the required black screw from more than 60 black screws of different sizes, lengths and postures at a speed of milliseconds? On August 11, the reporter learned from the Liangjiang Collaborative Innovation Zone that recently, the rowing boat technology on the shore of Mingyue Lake customized the first 3D vision-guided screw feeding system in the UHV industry for customers, achieving a new breakthrough.
What are the difficulties in breaking through? Take the aforementioned "grasping screws" problem as an example. The disorderly placed screws require high image recognition capabilities; to grab one screw from a pile of screws, the robot's grasping accuracy is required to be high; and the surface of the screw reflects a lot of light, causing light spots and glare when the industrial camera is imaging, which interferes with recognition accuracy and efficiency.
▲The robot is grabbing a black screw. Image source: provided by the company
Rowing Boat Technology chose to overcome these difficulties one by one.
First, we equip customers with high-precision hardware - 3D structured light cameras for accurate image recognition. Second, we use our self-developed high-precision hand-eye calibration algorithm to allow the robot arm and 3D camera to achieve tacit "hand-eye coordination" and find the required screws within 0.2 milliseconds.
Through the deep learning model, the algorithm can quickly find the required screws in the image. At the same time, by segmenting pixels, it can make the contour features of different screws clearer. Combined with posture calculation, it can calculate the position of the required screw in a pile of screws and grasp it accurately.
It is not enough to just know how to "grasp screws". When the R&D team was deeply exploring application scenarios, they found that when grabbing screws in an unordered manner, the robot arm is prone to collide with the material frame and cause damage. This not only slows down production, but also poses a huge safety hazard.
Rowing Boat Technology has added posture constraints and intelligent dodge functions to the robotic arm, which not only prevents the robotic arm from moving around, but also gives it the ability to "keep an eye on all directions" and automatically avoid obstacles when encountering them, ensuring production safety.
A relevant person in charge of Yaoluchuan Technology said that this technological breakthrough has solved the industry's difficult problem of disordered grasping in the field of machine vision. It has also opened up a new scenario for intelligent manufacturing with leading technologies such as full-field high-precision guidance, anti-glare, and anti-collision.
