Integrating YOLO and MiDaS for real-time distance measurement in flexible robotic systems

Abstract

With soft structure and high adaptability, flexible robotic systems can be widely applied in dynamic and complex environments, especially for tasks related to object recognition and interaction [1]. In order to have a more accurate measure of the object distance, this paper presents a distance measurement method that integrates YOLO (You Only Look Once) and DPT (Depth Prediction Transformer) technologies. It first uses the YOLO algorithm for object detection to extract the center coordinates of the object and then applies the DPT depth estimation model to recover depth information from the image [1]. It can efficiently calculate the actual distance between the target object and the camera by combining real-time detection capability with YOLO and depth estimation provided by DPT [2]. Compared to the traditional object detection or depth estimation method, the combined YOLO and DPT approach reveals great advantages in both aspects of measurement accuracy and real-time performance. Experimental results prove the effectiveness of the presented approach in object detection in dynamic environments and at the same time enhancing the perception capability of flexible robotic systems for performing object recognition and manipulation tasks with more precision in challenging scenes. Compared to the existing, traditional methods, our method is very promising in the field of robotic perception and interaction efficiency enhancement.