Spatial recognition and positioning system of robotic arm based on binocular vision

Abstract

Robotic arms are widely used in fields such as national defense construction, industrial manufacturing, and home services. However, there are many problems in the application of robotic arms in identifying and positioning target objects. At present, the identification and positioning technology of target objects by robotic arms are mostly designed independently and without interference. In order to solve this pain point, this article proposes a binocular vision system composed of two cameras to realize the identification and positioning of target objects. Its working principle is similar to how human eyes perceive the distance of objects. It uses two cameras to obtain environmental information, cooperates with the YOLOv8 target detection model's robotic arm spatial recognition and positioning system, and builds an experimental platform for experiments. The experimental results prove that this system can improve the accuracy of the robotic arm. The object recognition and precise grasping capabilities under the task of grasping small-volume objects are of significant significance to the improvement of the automation level of robotic arms in industry, agriculture, medical and other fields.