丘陵山地柑橘果园机器人自主导航与精确控制系统设计与试验  

Design and Test of Autonomous Navigation and Precise Control System for Robots Used in Citrus Orchards in Hilly and Mountainous Area

作　　者：刘杰[1] 付兴兰 李旭[1] 李川红 李光林[1] LIU Jie;FU Xinglan;LI Xu;LI Chuanhong;LI Guanglin(College of Engineering and Technology,Southwest University,Chongqing 400715,China)

机构地区：[1]西南大学工程技术学院,重庆400715

出　　处：《西南大学学报（自然科学版）》2024年第8期184-197,共14页Journal of Southwest University(Natural Science Edition)

基　　金：农业农村部gg项目(NK202216010402).

摘　　要：为实现丘陵山地柑橘果园机器人的自主导航,设计了一种基于激光雷达和惯性传感器的果园自主导航系统.使用激光SLAM融合多传感器所获数据,并利用误差状态卡尔曼滤波器进行位姿优化,从而构建果园的三维全局点云地图.对全局地图进行分析和处理,确定机器人的安全行驶区域,并应用基于机器人运动学多约束条件的三次非均匀B样条曲线轨迹优化算法实现路径规划,使用NDT-ICP算法实现机器人在全局地图中的定位.为适应丘陵山地果园地形复杂性,提出了一种基于预瞄跟踪的自校正增量PID控制策略,利用递归最小二乘法实时调整PID参数,并在机器人履带式行走机构上进行了系统整合测试.试验结果表明,误差状态卡尔曼滤波器优化后的地图精度与优化前相比有明显提升.机器人以1.2 m/s速度行驶时,导航控制系统的直线行驶平均位置偏差和航向角偏差分别为0.18 m和4.2°,转弯行驶平均位置偏差和航向角偏差分别为0.38 m和16.7°,可满足丘陵山地柑橘果园智能农机自主导航作业需求.To achieve autonomous navigation of robots in hilly and mountainous citrus orchards,a citrus orchard autonomous navigation system based on laser radar and inertial sensors was designed.The system utilized laser SLAM to fuse data from multiple sensors and an error state Kalman filter for pose optimization,thus constructing a three-dimensional global point cloud map of the orchard.The global map was then analyzed and processed to determine the robot s safe driving area.A trajectory optimization algorithm based on three-dimensional non-uniform B-spline curves and constraints of robot kinematics was applied to achieve path planning.The NDT-ICP algorithm was used for localization of robot in the global map.To adapt to the complexity of terrain in hilly and mountainous citrus orchards,a self-correction incremental PID control strategy based on preview tracking was proposed.The PID parameters were adjusted in real-time using recursive least squares method,and integration testing was conducted on the robot s track walking mechanism.The experimental results showed that the map accuracy was significantly improved after optimizing with the error state Kalman filter.When the robot traveled at a speed of 1.2 m/s,the average position error and heading angle error of the navigation control system in straight line driving were 0.18 m and 4.2°,respectively,while in turning driving,the average position error and heading angle error were 0.38 m and 16.7°,respectively.These results meet the requirements for autonomous navigation in smart agricultural machinery for hilly and mountainous citrus orchards.

关 键 词：柑橘果园机器人 激光SLAM 误差状态卡尔曼滤波器 NDT-ICP算法 自校正增量PID 

分 类 号：TN964[电子电信—信号与信息处理] S24[电子电信—信息与通信工程]