旱地株间除草机器人设计与试验

Design and Test of Interplant Weeding Robot in Dryland

摘要:

针对株间机械除草机器人株距适应性差、幼苗损伤率高等问题,本文设计一种基于视觉引导的智能株间除草机器人,由移动底盘、除草部件、视觉识别系统及控制系统构成。其中,除草机构采用非圆齿轮行星轮系传动,通过偏心轨迹规划实现避苗除草;视觉系统基于Otsu自适应阈值分割算法,建立基于面积阈值的作物-杂草识别定位;控制系统通过作物视觉识别结果控制除草机构驱动电机启停,实现株间作业轨迹控制。试验结果表明,机器人平均检测率为93.99%;当底盘前进速度为0.4 m/s时,不同工况下除草率不小于92.23%,整体平均除草率为93.70%,平均伤苗率为3.90%。试验结果验证了机器人对不同株距株间除草的适应性。

Abstract:

Due to the different distribution locations of farmland weeds,the overall can be divided into two categories: inter-row weeds,inter-plant weeding,in which inter-plant weeding is close to the crop,and the weeding operation is intermittent,so the key problem of mechanical weeding is to solve the complexity of inter-plant weeds. In order to solve the problems of poor plant spacing adaptability and high seedling damage rate of interplant mechanical weeding robot,an intelligent interplant weeding robot system was proposed based on visual guidance. The system consisted of a moving chassis,weeding parts,a visual identity system and a control system. Among them,the weeding mechanism adopted non-circular gear planetary gear train transmission,and realized seedling avoidance and weeding through eccentric trajectory planning. Based on the Otsu adaptive threshold segmentation algorithm,the vision system established crop-weed identification and positioning based on area threshold. The system used visual recognition results to control the movement trajectory of the weeding mechanism and the inter-plant operation time. The experiments showed that the robot's average recognition rate reached 93.99%;when the chassis moved at a speed of 0.4 m/s,the weeding rate was not less than 92.23% under different working conditions,with an overall average weeding rate of 93.70%,and the average seedling injury rate was 3.90%. The experimental results verified the adaptability of the robot for weeding between multiple crops.