基于高斯伪谱法的自动驾驶车辆状态研究

针对自动驾驶车辆的多阶段多约束轨迹优化问题,根据高斯伪谱法的思路,建立了基于场景的环境模型和车辆动力学模型,设置了车辆动力学约束、速度约束等各种状态约束,包括自动驾驶在每一阶段的起始状态和终了状态等参数的约束条件。采用高斯伪谱法通过将控制变量和状态变量进行离散化来获得其近似表达式,从而将自动驾驶的轨迹规划问题转化成性能指标的优化问题,最终求得了自动驾驶车辆的安全、有效的路径轨迹。研究结果表明:高斯伪谱法具有计算精度高、求解速度快的特点,能在考虑各种约束条件下,实现多阶段轨迹优化。

Study on the States of Autonomous VehicleBased on Gaussian Pseudospectral Method

Aiming at the problem of multi-stages multi constraints trajectory optimization of autonomous vehicle,based on Gaussian pseudospectral method,scenario-based environment model and vehicle dynamics model are established and a variety of state constraints covering vehicle dynamics constraints and speed constraint are defined,including the parameter constraint conditions of initial and final states in each stage of autonomous driving.Then Gaussian pseudospectral method is adopted to get the approximate expression by the discretization of control variables and state variables,hence the trajectory planning problem is transformed into the optimization problem of performance indicators and finally a safe and effective path trajectory of autonomous vehicle is obtained.The results of research show that Gaussian pseudospectral method has the features of high computation accuracy and fast solving speed,and can achieve multi-stages trajectory optimization with consideration of various constraint conditions.