基于整车在环仿真的自动驾驶汽车室内快速测试平台

整车在环仿真测试方法可以安全、高效地验证复杂环境和极端工况等场景下自动驾驶汽车性能的有效性，基于此研发一种基于整车在环仿真的自动驾驶汽车室内快速测试平台，该平台由前轴可旋转式转鼓试验台、试验台测控子系统、虚拟场景自动生成子系统、虚拟传感器模拟子系统、驾驶模拟器、自动驾驶汽车和测试结果自动分析评价子系统组成。通过在试验台滚筒上独立加载转矩模拟车辆行驶阻力，可动态模拟不同的路面附着系数，同时利用坡度、侧倾和转向随动机构可模拟车辆俯仰角、侧倾角和航向角3个自由度；采用虚拟现实技术柔性集成车辆动力学模型、传感器仿真、复杂道路交通环境及测试用例仿真，模拟多种道路交通场景，并通过传感器仿真及数据融合等技术快速测试自动驾驶汽车智能感知与行为决策等性能指标。将自动驾驶汽车、虚拟仿真场景和试验台耦合构建一个闭环系统，完成了多项关键技术研发，包括：多自由度高动态试验台结构设计、虚拟测试场景自动重构方法和传感器数据模拟及注入方法，可满足在各种场景下测试自动驾驶汽车整车性能的需求。此外，为验证快速测试平台的有效性，以U-turn轨迹跟踪控制为研究实例，基于简化的车辆运动学模型和模型预测控制算法，在平台上搭建U-turn场景并对自动驾驶汽车的轨迹跟踪控制算法性能进行大量测试。结果表明：自动驾驶汽车室内快速测试平台可以真实地模拟汽车在道路上的运行工况，自动驾驶汽车在虚拟场景中的轨迹跟踪效果良好，与参考轨迹的偏差小于8%，证明了该测试平台检测方法的有效性。

An Indoor Rapid-testing Platform for Autonomous Vehicle Based on Vehicle-in-the-loop Simulation

The vehicle-in-the-loop simulation test method can validate the performance of autonomous vehicles safely and efficiently in complex environments and extreme conditions. This paper develops an indoor rapid-testing platform for autonomous vehicles based on vehicle-in-the-loop. This platform is comprised of seven subsystems:front-axle rotatable drum bench, automobile test bench detection system, virtual scenario automatic generation subsystem, virtual sensor simulation subsystem, driving simulators, autonomous vehicle, and result analyzing and evaluating automatically subsystem. The driving resistance was simulated by independently loading torque on the rollers supporting the four wheels of the automobile, so that different road adhesion coefficients were simulated. Slope, roll, and yaw follower mechanisms were used to simulate three degrees-of-freedom such as the pitch angle, roll angle, and course angle. Virtual reality technology was adopted to simulate various road traffic scenarios to verify performance indicators, such as intelligent perception and decision-making behavior of the autonomous vehicle, by means of flexible integration of the vehicle dynamics model, sensor simulators, simulations of complex traffic environments, and test cases. A closed loop system was modeled by coupling the automobile, the virtual simulation scenarios and the test bench in order to research and develop a number of key technologies, including the structural design of multi degrees-of-freedom and high-dynamic detection bench, automatic reconstruction method of virtual test scenes, and methods of simulation and injection for sensor data, which can meet the need to test the performance of autonomous vehicles in various scenarios. Further, to verify the effectiveness of the test platform, U-turn trajectory tracking control was taken as a research example. Based on a simplified vehicle kinematics model and the Model Predictive Control algorithm, a large amount of experiments was performed in virtual scenarios of U-turns to test the efficiency of the trajectory tracking control algorithm on the autonomous vehicle. The results showed that the platform could simulate automobile driving conditions on the road realistically, and the trajectory tracking effect of the autonomous vehicle in the virtual scenario is satisfactory. The deviation from the predicted trajectory is less than 8%, which demonstrates the effectiveness of the test platform.