人机协作系统中车辆轨迹规划与轨迹跟踪控制研究

自动驾驶系统需具备响应驾驶人意图且有效执行驾驶人意图的能力，以解决人机协作系统中存在的人机冲突、人机优势融合等问题。提出决策层“以人为主”、执行层“以机为首”的人机协作关系，构建包含驾驶人意图识别模块、基于意图识别的轨迹规划模块与轨迹跟踪控制模块的人机协作一体化控制系统框架，并重点对轨迹规划模块与轨迹跟踪控制模块开展研究。首先，结合双向长短期记忆神经网络（Bi-directional Long Short Term Memory，Bi-LSTM）与注意力机制模型建立换道轨迹规划模型；在改进人工势场算法中引入模型预测控制并建立避险轨迹规划模型。其次，通过开展驾驶模拟器试验建立换道与避险驾驶行为数据集，为拟人化模型训练和模型参数确定提供支撑。然后，综合考虑车辆状态变量、控制输入与输出以及道路结构参数等约束条件，构建基于最优转向前轮输入的线性时变模型预测轨迹跟踪控制器，实现对规划轨迹的精准跟踪。最后，基于驾驶模拟器搭建人机协作系统硬件在环测试平台，对轨迹规划模块与轨迹跟踪控制模块开展硬件在环测试与验证。结果表明：换道与避险规划轨迹光滑且平稳，轨迹跟踪控制过程中，车辆航向角与前轮转角变化平稳；所构建的轨迹规划与轨迹跟踪控制模块在确保安全性前提下可实现不同场景中的车辆运动控制需求。

Vehicle Trajectory-planning and Trajectory-tracking Control in Human-autonomous Collaboration System

An autonomous driving system should effectively respond to and carry out driver intentions so that human-machine conflicts may be resolved, and advantage integration may be achieved in a human-autonomous collaboration system. A novel human-autonomous collaborative relationship that is "human-oriented" in the decision-making layer, and "machine-oriented" in the execution layer was proposed, and a perceptual-decision-execution integrated human-autonomous collaboration control framework including a driver intention recognition module, trajectory-planning module, and trajectory tracking control module was established. This study focused on the trajectory-planning module and trajectory-tracking control module in a human-autonomous collaboration system. First, a lane change trajectory-planning model was constructed by combining bidirectional long short-term memory (Bi-LSTM) with an attention mechanism model. An improved artificial potential field (APF) algorithm with model prediction control (MPC) was employed to establish an obstacle-avoidance trajectory-planning model. Second, a driving simulator was used to conduct characteristic analysis experiments of driving behaviors, providing support for the training of a human-like model and the determination of model parameters. Third, a linear time-varying MPC trajectory tracking controller that considers the constraints of vehicle state variables, control inputs and outputs, and road-structure parameters was developed to track the local obstacle avoidance trajectory. Finally, based on the driving simulator, a hardware-in-the-loop test platform of the human-autonomous collaboration system was constructed, and verification tests were carried out for the trajectory-planning and the trajectory-tracking control module. The test results indicate that the lane-change planning trajectory and avoidance planning trajectory are smooth and stable. During the trajectory-tracking control process, the vehicle heading angle and front-wheel angle change smoothly. The trajectory-planning and the trajectory-tracking control module in the human-autonomous collaboration system can meet vehicle movement control requirements in different scenarios under the premise of ensuring driving safety.