Research And Implementation Of The Fusion Of Vision And Tactile In Robot Grasping

In recent years,with the rapid development of artificial intelligence and intelligent robot technology,machine vision and natural interaction robots have become the current research hotspot.In the field of robotic intelligent interaction,the robot's target recognition,locating,and grasping problems are one of the problems that must be solved when robots enter the smart society.The robot's grasping of target object mainly includes the following two parts:Firstly,the target object's information is taken to identify the target and determine the three-dimensional coordinates;secondly,the robot sense the information of object,improves the success rate of grasp.Today,robotic grasping algorithms:On the one hand,the real-time character of-target recognition and positioning is not good,or the recognition algorithm cannot perform three-position positioning.On the other hand,there is no tactile judgment when the object is grasped,which may cause the object to slide or be damaged when the object is grasped.In order to solve these problems,this dissertation focuses on the study of the fusion of vision and tactile of service robots,and designs a visual and tactile fusion system for robot grabbing,and validates the effectiveness of its fusion results.Research is mainly divided into three aspects:(1)In order to solve the problem of the robot's positioning to the target object,based on the YOLO(You Only Look Once)v2 object recognition algorithm,a new algorithm YRLA(YOLO-based Recognition and Location Algorithm)is proposed,which can achieve real-time target recognition and target positioning for target objects that need to be grasp.(2)In order to judge the robot's sense of force,a tactile feedback circuit was designed to extract and process the robot's judgment of force when gripping the object.(3)In order to improve the robot's success rate of grasp,the integration of tactile sensation and vision has greatly increased the success rate of the robot's grabbing of objects.The main contribution of this work has two points:the first is propose the YRLA algorithm so that the robot can reach a positioning accuracy of 0.2 cm to 1 cm within one meter;the second is to fuse the robot's vision and tactile,and the robot grasp successfully.The rate has increased by four percentage points.