Research On Navigation And Control Of Agricultural Robot Based On Visual Servo Feedback

With the acceleration of agricultural modernization in China,agricultural production is developing towards intellectualization and accuracy.The research on agricultural robots is getting hotter and hotter.This paper mainly studies the problem of automatic identification of agricultural robots in the field,and controls the path tracking of robots.It includes two aspects.1.Path recognition based on vision.The vision based path recognition of agricultural robot has many problems,such as low recognition rate,slow speed,easy exposure to light,and influence of weeds.In order to solve these problems,this paper uses light independent quantity to eliminate the illumination interference.The mixed threshold method is used to segment the image,and the recognition speed is speeded up on the basis of improving the recognition rate.The recognition of crop rows is carried out by projection method,and the line method of migration line is used to divide the crop rows to improve the recognition rate.Finally,the linear regression of crop rows is carried out by using the least square method to speed up the crop row recognition speed.When the agricultural robot navigate automatically,the position of the camera is basically fixed.In this paper,the image coordinate of crop row is transformed into the world coordinate to get the offset between the agricultural robot in the world coordinate system and the navigation path.Compared with the traditional crop row recognition method of agricultural robot,this method has the advantages of small data computation short time consuming and high accuracy,which can effectively determine the navigation path of agricultural robot.2.The research of path tracking based on reinforcement learning method.In this paper,based on the kinematic modeling and analysis of the wheeled agricultural robot,the results of image processing are used as feedback information,and the method of enhanced learning compensation PID is used to make a double wheel speed decision.The DC motor is used in the motor speed control.Based on the modeling and analysis of the DC motor,the PI controller is used to control the speed of the DC motor with the encoder information.It is proved by experiments that the enhanced learning compensation PID based on Actor-Critic has excellent effect in path tracking,and can effectively correct the path tracking error of agricultural robot.