Safety-Critical Cooperative Target Enclosing Control of Autonomous Surface Vehicles Based on Finite-Time Fuzzy Predictors and Input-to-State Safe High-Order Control Barrier Functions
Safety-Critical Cooperative Target Enclosing Control of Autonomous Surface Vehicles Based on Finite-Time Fuzzy Predictors and Input-to-State Safe High-Order Control Barrier Functions
复制标题
DOI：
10.1109/tfuzz.2023.3309706
复制
发表时间：
2024-03
影响因子：
11.9
通讯作者：
Yue Jiang;Zhouhua Peng;Lu Liu;Dan Wang;Fumin Zhang
中科院分区：
文献类型：
--
作者：
Yue Jiang;Zhouhua Peng;Lu Liu;Dan Wang;Fumin Zhang
本文研究了欠驱动自动水面车辆（asv）在障碍物作用下的协同目标封闭问题。除了模型非线性引起的未知动力学、未知输入增益和外部干扰外，每个ASV还受到输入约束的影响。提出了一种机动目标车辆的安全关键型协同目标包围控制方法。具体而言，提出了一种有限时间模糊预测器，利用历史车辆数据的积分来学习未知动力学。利用分布式目标估计器恢复目标位置，建立了标称分布式目标封闭控制律，实现了绕航编队。为了避免自动驾驶汽车与障碍物/团队成员之间的碰撞，首先引入了输入到状态安全的高阶控制屏障函数来编码安全约束。基于安全约束和输入约束，构造了一个二次规划问题，利用投影神经网络对最优解进行跟踪，得到了最优安全临界控制律。通过李雅普诺夫理论证明了闭环控制系统的输入到状态稳定。此外，无论高阶相对度如何，多个ASV系统都被证明是安全的。该方法的突出贡献在于有限时间模糊学习和干扰下的无碰撞目标封闭控制。仿真结果验证了所提出的无模型安全关键控制方法对机动目标协同包围的有效性。
This article addresses cooperative target enclosing of underactuated autonomous surface vehicles (ASVs) subject to obstacles. Each ASV suffers from input constraints in addition to unknown kinetics induced by model nonlinearities, unknown input gains, and external disturbances. A safety-critical cooperative target enclosing control method is proposed for surrounding a maneuvering target vehicle. Specifically, a finite-time fuzzy predictor is presented to learn the unknown kinetics with the integral of historical vehicle data. By using a distributed target estimator to recover the target position, a nominal distributed target enclosing control law is developed to achieve a circumnavigation formation. To avoid collisions between ASVs and obstacles/team members, input-to-state safe high-order control barrier functions are first introduced for encoding safety constraints. Based on the safety constraints and input constraints, a quadratic programming problem is formulated, and an optimal safety-critical control law is obtained by using projection neural networks to track the optimal solution. The closed-loop control system is proven to be input-to-state stable via Lyapunov theory. Moreover, the multiple ASV systems are proven to be input-to-state safe regardless of high-order relative degree. The salient contributions of the proposed approach lie in finite-time fuzzy learning and collision-free target enclosing control under disturbances. Simulation results validate the effectiveness of the proposed safety-critical model-free control method for cooperatively surrounding a maneuvering target.