Environment Perception And Motion Planning For Intelligent Industrial Robot

Intelligent robots can perform specific tasks autonomously,and research on environmental perception and motion planning can help to increase the autonomy of robots and make them more intelligent.The environment perception obtains point cloud or image data through vision sensors,then this data is analyzed and processed by a computer,so that the robot system can obtain various kinds of information of objects in the environment.The motion planning is based on the environment perception,and the movement of the robot can be controlled by planning a non-collision trajectory.This paper first introduces the calibration methods of various types of sensors to achieve the unification of different data in the coordinate system.Based on the analysis of the existing calibration algorithm,the Navy hand-eye calibration method is used to calculate the least squares solution,so that the calibration accuracy reaches millimeter level;and the single-line laser sensor calibration method based on the two-step method can guarantee that the accuracy is sufficient,and effectively reduce the amount of calculation,thereby improving the efficiency of the algorithm.As a commonly used surface description method for three-dimensional objects,point cloud has obvious advantages in environmental perception.In this paper,region growing,RASANC,cluster analys is and other point cloud segmentation algorithms are used to identify and segment objects in the robot workspace,and their size and pose are calculated by constructing a directed bounding box.In addition,the high-precision point cloud obtained by the laser sensor can also reconstruct the surface model of the object.The image is one dimension less than the point cloud,so it is more convenient in data processing,and the convolut ional neural network makes the image more widely used.In order to complete the robot's automatic grasp task,this paper designs a seven-layer neural network to quickly identify the grab points of objects in the image.In addition,the MobileNet-based SSD network is also used for the identification and location of mult i-target objects,and the fina l recognition rate can reach 80%,and the network running speed is only 17 ms.The motion planning of robots is the basis for robots to perform various tasks.The traditional teaching planning methods have many defects,so this paper designs an independent planning scheme.First,collision detection will determine the free-form space of the robot,and the use of the PRM algorithm will plan a series of path points in that space.Finally,a multi-point planning method is used to plan a collision-free motion trajectory from the path point.This planning scheme allows the robot to effectively avoid obstacles and reduce collision damage while exercising.In a word,based on the existing methods,this paper designs a set of robotic environment perception and motion planning methods,which has the characteristics of good stability and high algo rithm efficiency.At the same time,the algorithm proposed in this paper can basically let the robot autonomously complete tasks such as scanning and grabbing.