基于Huber函数非线性模型预测控制的自主水下机器人自适应轨迹跟踪控制方法

Adaptive Trajectory Tracking Control of AUV Based on Huber-function-focused Nonlinear Model Predictive Control

摘要:

自主水下机器人(AUV)的稳定与高精度控制面临环境的外部扰动、内部动态耦合干扰等困难,为了解决AUV在复杂扰动下的平面轨迹跟踪问题,本文提出了一种基于Huber函数的非线性模型预测控制(HNMPC)和自适应扩张状态观测器(AESO)的控制算法。针对突变扰动的问题,提出了HNMPC方法,Huber函数提高了稳态时的鲁棒性。采用扩张状态观测器处理AUV未知动力学建模和未知扰动的问题;针对扩张状态观测器的峰值现象和振荡问题,提出了自适应扩张状态观测器,减少了初始状态的峰值并提高了稳态精度。本文提出的算法在Gazebo仿真平台进行验证,实验表明,在模拟洋流干扰下的正弦轨迹跟踪任务中,本文提出的算法比NMPC + ESO算法在纵荡、横荡和偏航方向上的积分绝对误差分别减少了41%、35.09%、40.78%,并通过与PID算法和NMPC算法比较体现了算法的优势。最后搭建了水池实验方案,所提出的算法比NMPC + ESO算法在纵荡、横荡和偏航方向上的积分绝对误差分别减少了41.22%、35.54%、0.79%,说明了本文算法在实际环境中的有效性。

Abstract:

Achieving stable and high-precision motion control of autonomous underwater vehicles ( AUVs) faces severe challenges such as complex external environmental disturbances and internal dynamic coupling interference. To solve the horizontal planar trajectory tracking problem of AUVs under these complex disturbances, a novel joint control algorithm was proposed based on Huber-function-focused nonlinear model predictive control (HNMPC) and an adaptive extended state observer (AESO). Addressing the critical issue of sudden mutational disturbances, the HNMPC method was developed, where the introduced Huber penalty function effectively enhanced the system robustness at steady state. Meanwhile, an extended state observer (ESO) was employed to actively handle the unmodeled dynamics and unknown lumped disturbances of the AUV. To effectively mitigate the peaking phenomenon and high-frequency oscillations inherent in traditional ESOs, the AESO was proposed, which successfully reduced the initial state transient peaks and improved steady-state control precision. The performance of the proposed algorithm was firstly validated within the Gazebo simulation platform. Simulation results in a complex sinusoidal trajectory tracking task under simulated ocean current disturbances demonstrated that the proposed algorithm reduced the integral absolute error (IAE) in the surge, sway, and yaw directions by 41%, 35.09%, and 40.78%, respectively, compared with the NMPC + ESO algorithm; its advantages were further demonstrated through comparative evaluations with traditional PID and standard NMPC algorithms. Finally, a physical pool experiment platform was established for real-world validation. The experimental results showed that the proposed algorithm reduced the IAE in the surge, sway, and yaw directions by 41.22%, 35.54%, and 0.79%, respectively, compared with the NMPC + ESO algorithm, thoroughly validating the effectiveness and practical applicability of the proposed method in real-world environments.