基于间接共享控制的智能车协同接管方法

Cooperative Takeover Method for Automated Vehicles Based on Indirect Shared Control

从自动驾驶模式向手动驾驶模式切换时的控制权移交问题已成为智能车领域的研究热点。为了改善驾驶人从非驾驶任务状态接管车辆时因认知和操作能力不足而引起的安全性和稳定性问题,在总结现有研究成果的基础上,提出一种基于间接共享控制的智能车协同接管控制方法。首先,定义协同接管方法的应用对象是控制系统良好、全局感知或决策受损需要接管的车辆,接管场景是时间较为紧迫的前方冲突;接着,采用Sigmoid函数设计局部路径规划算法,在满足无碰撞和横向稳定性约束条件的基础上,以曲线斜率最小化为优化目标规划出系统控制的目标轨迹;然后,建立车辆动力学和运动学模型,采用模型预测控制算法,以跟踪目标路径的误差最小化和转向角最小化为优化目标,计算自动系统的控制输出量;同时,采用碰撞时间倒数和横向加速度作为车辆纵、横向行驶风险的表征指标,设计可根据行驶风险动态调整的控制权重分配模型,实现控制权平稳过渡;最后,以真实驾驶人作为被试,模拟自动驾驶条件下驾驶人的非驾驶任务状态,通过模拟驾驶试验验证所设计的协同控制方法对驾驶人接管过程的安全性、稳定性绩效,以及驾驶负荷等生理指标的改善效果。研究结果表明:该协同控制方法能显著提高接管后的安全性和稳定性绩效,改善驾驶人在接管过程中的负荷突变现象,使得系统接受度得到显著提升。

The transfer of control from automatic to manual driving has become a popular research topic in the field of intelligent vehicles.In this study,an intelligent vehicle cooperative takeover method based on indirect shared control is proposed to improve the safety and stability problems caused by the lack of cognition and operational ability when the driver takes over control of the vehicle from non-driving related tasks based on a summary of previous studies.First,it was determined that the application scenario of the cooperative takeover method is the vehicle that needs to be taken over with a damaged global perception but good control resources.The takeover scenario is forward obstacle avoidance with tight time.Second,a local path-planning algorithm was designed using a sigmoid function.On the basis of meeting the constraints of collision-free and lateral stability,the target trajectory was planned with minimization of the curve slope as the optimization objective.Third,vehicle dynamics and kinematics models were established.A model predictive control algorithm was adopted to minimize the error in tracking the target trajectory and steering angle,and the control output of the automatic system was calculated.Simultaneously,the reciprocal of time-to-collision and lateral acceleration were utilized as the characterization indices of vehicle longitudinal and lateral driving risk,and a control weight distribution model that can be dynamically adjusted according to driving risk was designed to facilitate a smooth control transition.Finally,real drivers were invited as participants,and their non-driving-related task states under automated driving conditions were simulated.The driving simulator experiment verified the improvement effect of the designed cooperative takeover method on the safety,stability performance,workload,and other physiological indicators of the driver takeover process.Based on the results,it can be stated that the cooperative takeover method could significantly improve safety and stability performance after takeover,improve the workload mutation phenomenon of drivers,and significantly improve system acceptance.