农业机器人视觉导航的预测跟踪控制方法研究

Predictive tracking and control method for vision-guided navigation of agricultural robot

农用拖拉机的视觉导航技术可以帮助人员远离某些高温、高湿以及有毒害的作业环境,提高作业的自动化智能化程度,还能实现精确定点作业以促进农业可持续发展等。该文首先分析了轮式拖拉机跟踪引导路径的行为特点,建立起相应的非线性随机数学模型。而后,基于卡尔曼滤波的思想融合了各传感器的观测值给出预测跟踪控制方法。避免了视觉系统为主的计算耗时导致状态反馈滞后而产生的不利影响,改善了导航控制的鲁棒性和精度。仿真和初步试验结果都表明了此方法的有效性。

To avoid the unfavorable working conditions and guarantee the sustainability of agriculture, it is widely thought that the wheeled mobile robot guided by machine vision, substituting for conventional tractors in some farming activities, will play an important role in the future. The time lag, however, produced mainly by robot vision system and the other signal processing will exert some negative effects on the autonomous navigation of the wheeled mobile robot. Based on Kalman filtering, the measurements of all sensors were fused and a predictive control method was developed skillfully to overcome it. Then the kinematical behavior of the wheeled mobile robot was analyzed in detail, and the corresponding nonlinear stochastic model equation and the observation equation were set up respectively. Both the simulation and the initial experiment show that this method is effective for the visual navigation of the wheeled mobile robot in the field.