Distributed Model Predictive Control for Connected and Automated Vehicles in the Presence of Uncertainty
Distributed Model Predictive Control for Connected and Automated Vehicles in the Presence of Uncertainty
复制标题
存在不确定性的联网和自动驾驶车辆的分布式模型预测控制
DOI：
10.1115/1.4054696
复制
发表时间：
2022
期刊：
影响因子：
--
通讯作者：
Bhattacharyya, Viranjan
中科院分区：
文献类型：
--
作者：
HomChaudhuri, Baisravan;Bhattacharyya, Viranjan
本文重点介绍了分布式鲁棒模型预测控制（MPC）方法的发展，多个连接和自动化车辆（CAV），以确保其安全运行的不确定性。提出的分层控制框架包括参考轨迹生成、分布式鲁棒障碍占用集计算、分布式状态约束集评估、数据驱动线性模型表示和基于管的鲁棒MPC设计。为了实现CAV之间的分布式操作，我们提出了一种方法，该方法利用基于采样的参考轨迹生成和分布式约束集评估方法，该方法将CAV之间的耦合避碰约束进行了简化。其次是数据驱动的线性模型表示的非线性系统，以评估凸等效的非线性控制问题。最后，为了确保在存在不确定性的情况下安全运行，本文采用了一种鲁棒的基于管的MPC方法。对于多CAV换道问题，仿真结果表明，所提出的控制器的计算效率和能力，以分布式的方式产生安全和平滑的CAV轨迹方面的功效。
This article focuses on the development of distributed robust model predictive control (MPC) methods for multiple connected and automated vehicles (CAVs) to ensure their safe operation in the presence of uncertainty. The proposed layered control framework includes reference trajectory generation, distributionally robust obstacle occupancy set computation, distributed state constraint set evaluation, data-driven linear model representation, and robust tube-based MPC design. To enable distributed operation among the CAVs, we present a method, which exploits sampling-based reference trajectory generation and distributed constraint set evaluation methods, that decouples the coupled collision avoidance constraint among the CAVs. This is followed by data-driven linear model representation of the nonlinear system to evaluate the convex equivalent of the nonlinear control problem. Finally, to ensure safe operation in the presence of uncertainty, this article employs a robust tube-based MPC method. For a multiple CAV lane change problem, simulation results show the efficacy of the proposed controller in terms of computational efficiency and the ability to generate safe and smooth CAV trajectories in a distributed fashion.