Path following control of the automatic guided vehicle (AGV) for material handling in the plant requires a smooth and high-precision continuous attitude adjustment. However, there are many difficulties in designing the controller, such as actuator saturation, parameter selection, and the influence of deviations relations. In this paper, an improved model predictive control SCMPC-ST with a state classification model (SCM) and a smooth transition (ST) strategy is proposed to address these problems. First, based on the deviations relations and the actuator saturation, the SCM is designed to divide the pose states of AGV into three stages. With clearly objective functions and boundary constraints, SCM allows the pose states can be transferred sequentially, which avoids the problem of parameter selection. Second, use the analytical method to resolve SCM to directly obtain the desired control law, which provides excellent performance in real-time. Finally, the smooth transition strategy is built to adjust the control law at the transition step and overshoot step to ensure the stability of the control process. Simulation results show that by focusing on adjusting the deviations relations, SCMPC-ST can make AGV eliminate all deviations continuously and smoothly while maintaining high accuracy.

用于工厂物料搬运的自动导引车 (AGV) 的路径跟踪控制需要平稳且高精度的连续姿态调整。然而，控制器的设计存在很多困难，如执行器饱和、参数选择、偏差关系的影响等。在本文中，提出了一种具有状态分类模型 (SCM) 和平滑过渡 (ST) 策略的改进模型预测控制 SCMPC-ST 来解决这些问题。首先，基于偏差关系和执行器饱和度，SCM 被设计为将 AGV 的位姿状态分为三个阶段。有了明确的目标函数和边界约束，SCM 允许位姿状态可以顺序转移，避免了参数选择的问题。第二，使用解析法解析SCM，直接得到所需的控制律，实时性优异。最后构建平滑过渡策略，在过渡步和超调步调整控制律，保证控制过程的稳定性。仿真结果表明，SCMPC-ST通过专注于调整偏差关系，可以使AGV在保持高精度的同时，连续、平稳地消除所有偏差。