本文在描述车辆动力学的车辆数学模型基础上，提出并解释了自主驾驶的分层概念。第一级是路径规划，在起点和终点之间寻找一条路径。然后根据找到的路径和一个简单的运动学车辆模型，找到下一层次概念的初始输入，即路径优化。在路径优化层面，基于给定的目标函数找到一条最优路径。最优路径是车辆相对于不同目标（如能耗、舒适性和安全性）的轨迹。然后假设车辆上使用的传感器提供道路上最终障碍物的位置、速度和方向信息，更新路径优化水平，以寻找到一条无碰撞路径。由于几乎不可能有一个考虑所有不确定性、外部因素和干扰的精确模型，因此增加了一个新的层次，即路径控制，它产生额外的输入，以补偿最佳状态和车辆对最优输入响应之间的差异。

在本文中，每一个层次的自主驾驶概念都会有更详细的解释。为了展示概念的正确功能，它们在不同的软件（如Matlab和IPGCarMaker）中进行了建模、实现和仿真，并在马格德堡奥托冯盖瑞克大学（Otto von Guericke University of Magdeburg）名为BugEE的小型车辆上实现。

In this thesis, based on vehicle mathematic model which describes its dynamics, a hierarchical concept for autonomous driving is developed and explained. First level is path planning which finds a path between start point and end desired point. Then based on found path and a simple kinematic vehicle model initial inputs for next level of hierarchical concept is found which is path optimization. On path optimization level, based on a given objective function an optimal path is found. An optimal path is a trajectory for the vehicle with respect to different objectives such as energy consumption, comfort and safety. Then supposing sensors used in vehicle give us information about position, velocity and orientation of eventual obstacles on the road, path optimization level is updated in order to find a collision free path. Forasmuch as having an exact model which considers all uncertainty, external factors and disturbances is nearly impossible, a new level named path control is added which generates additional inputs to compensate the difference between optimal states and vehicle response to optimal inputs. Each of these levels of presented hierarchical concept of autonomous driving are explained with more details in this thesis. To show the correct functionality of the concept, they are modeled, implemented and simulated in different software such as Matlab and IPGCarMaker and also implemented on a small size vehicle build at Otto-von-Guericke University of Magdeburg named BugEE.

1.       引言

2. 轮胎与车辆动力学

3. 路径规划策略

4. 路径优化等级

5. 障碍物避撞

6. 路径控制与仿真结果

7. 实验设置与结果

8. 结论与展望

更多精彩文章请关注公众号：