基于复合终端滑模的无人农机自主导航系统设计与试验

Design and Experiment of Autonomous Navigation System for Unmanned Agricultural Machinery Based on Composite Terminal Sliding Mode

摘要:

无人农机自主导航系统在推动智慧农业的发展中扮演着至关重要的角色。 本文设计了一种基于机器人操作系统 2(Robot operating system 2,ROS2)的无人农机自主导航系统,旨在满足多种典型农业场景的精准作业需求。 该导航系统由 3 个核心部分组成:具有模块化设计的无人农机硬件平台,基于多传感器融合的环境感知与路径规划系统,以及采用基于模型设计方法的核心控制系统。 设计了一种基于时变参数扩张状态观测器(Extended state observer,ESO)的固定时间复合终端滑模控制方法用于核心控制部分。 首先,时变 ESO 能够实时估计农田中的外部干扰,且抑制传统 ESO 的尖峰现象。 其次,设计积分滑模面来抑制传统终端滑模中的抖振,并在此基础上得到复合终端滑模控制器。 通过仿真分析,验证了所设计的控制算法具有较好的稳定性和动态响应性能。 更进一步,对东风 1204-CVT 型拖拉机进行了电动化改造,并基于所设计的无人农机自主导航系统分别进行低速和高速工况下的田间施药作业实验。 最后,实验结果表明,以速度 3 km / h 作业时,直线跟踪横向偏差的最大绝对误差为 0. 246 m, 平均绝对误差为 0. 068 m,标准差为 0. 049 m;转弯跟踪横向偏差绝对值的最大值为 0. 279 m,平均值为 0. 099 m,标准差为 0. 069 m;以速度 6 km / h 作业时,直线跟踪横向偏差的最大绝对误差为 0. 400 m,平均绝对误差为 0. 121 m, 标准差为 0. 087 m;转弯跟踪横向偏差绝对值的最大值为 0. 488 m,平均值为 0. 158 m,标准差为 0. 101 m。 综上,所设计的复合终端滑模控制方法在路径跟踪精度方面具有显著优势。

Abstract:

Autonomous navigation system for unmanned agricultural machinery plays a crucial role in promoting the development of smart agriculture. A precise autonomous navigation system based on robot operating system 2 (ROS2 )was developed, aiming to meet the demand for precision operations in a variety of typical agricultural scenarios. The navigation system consisted of three core components: an unmanned agricultural machinery hardware platform with a modularized design, an environment sensing and path planning system based on multi-sensor fusion, and a core control system based on the model based design (MBD)method. A fixed-time composite terminal sliding mode control method based on the time-varying parameter extended state observer (ESO)was designed for the core control part. The time- varying ESO can estimate the external disturbances in the farmland environment in real-time and effectively suppress the peaking phenomenon of the traditional ESO. By designing an integral sliding mode surface, the chattering of the traditional terminal sliding mode was suppressed, and then a composite terminal sliding mode controller was constructed to generate smooth path tracking control signals, achieving stable control of the unmanned agricultural machinery. Through simulation analysis, the designed control algorithm was verified to have good stability and dynamic response performance. In order to further verify the practical application of the system, the Dongfeng 1204-CVT tractor was electrified to carry out field spraying operation experiments under both low and high speed conditions. The experimental results showed that when traveling at speed of 3 km / h, the maximum absolute error of the lateral offset in straight-line tracking was 0. 246 m, the mean absolute error was 0. 068 m, and the standard deviation was 0. 049 m; the maximum absolute error of the lateral offset in turning tracking was 0. 279 m, the mean absolute error was 0. 099 m, and the standard deviation was 0. 069 m. When operating at higher speed of 6 km / h, the maximum absolute error of straight line tracking lateral offset was 0. 400 m, the average absolute error was 0. 121 m, and the standard error was 0. 087 m; the maximum absolute error of the lateral offset in turning tracking was 0. 488 m, the mean absolute error was 0. 158 m, and the standard deviation was 0. 101 m. In summary, the designed fixed-time composite terminal sliding-mode control method adopted had significant advantages in terms of the path tracking accuracy.