基于ROS的Autoware开源项目

Autoware学习（一）——初步了解

Autoware(https://github.com/autowarefoundation/autoware)是世界上第一款用于自动驾驶汽车的“一体化”开源软件。 Autoware的功能主要适用于城市，但也可以涵盖高速公路，高速公路，中间地区和地理围栏。 Autoware的代码库受Apache 2许可证保护。 请自行决定使用它。 为了安全使用，我们为那些没有真正自动驾驶汽车的人提供基于ROSBAG的仿真环境。 如果您计划将Autoware与真实的自动驾驶汽车一起使用，请在现场测试前制定安全措施和风险评估。其支持以下功能：

3D本地化
3D映射
路径规划
路径跟随
加速/制动/转向控制
数据记录
汽车/行人/物体检测
交通信号检测
交通灯识别
车道检测
对象跟踪
传感器校准
传感器融合
面向云的地图
连接自动化
智能手机导航
软件仿真
虚拟现实
　　Autoware是基于ROS的开源软件，可以在开放的城市区域部署自动驾驶移动性。 它提供但不限于以下模块。 通过3D地图和SLAM算法结合GNSS和IMU传感器实现定位。 检测使用具有传感器融合算法和深度神经网络的相机和LiDAR。 预测和规划基于概率机器人和基于规则的系统，部分使用深度神经网络。 Autoware对车辆的输出是速度和角速度。 这是Control的一部分，虽然Control的主要部分应该驻留在车辆的线控控制器中，其中经常采用PID和MPC算法。

To Begin With
Build a pointcloud map
Autoware provides a 3D mapping module based on simultaneous localization and mapping (SLAM). It builds a local-coordinate pointcloud map while manually driving the vehicle. You should be careful about a loop closure constraint. If you drive too far from the origin, the resulting pointcloud map might be misaligned with the real geographical features, unless you use the Graph SLAM method. That is, the pointcloud map built by Autoware alone should not be large-scale.

Make a route in the pointcloud map
Once the pointcloud map is ready, you can make a route that you want to follow autonomously. The most straightforward way to make a route is that you drive manually once on the route and record vehicle positions on a time. This trace of vehicle positions can be saved as waypoints. All you have to do is to follow these waypoints. A more intelligent way, on the other hand, is that you use VectorMapper to add road features to the pointcloud map, and follow the waypoints comprising the center line of the lane of your route; however this is a time-consuming task.

Follow waypoints of the route
Once you make a route with waypoints, all you have to do is just to follow these waypoints, controlling the speed/throttle and the angle/steering based on the localization and detection results. To go beyond, for example on a public road, you would need to activate more modules, such as the object tracker, traffic light recognizer, and decision maker. You also suppose that, in practice, the waypoints is generated by some high-accuracy navigation service, which will be supported on Autoware.AI. Once the waypoints are ready, the exercise you try herein will be useful in any test field.

Research Papers for Citation
S. Kato, S. Tokunaga, Y. Maruyama, S. Maeda, M. Hirabayashi, Y. Kitsukawa, A. Monrroy, T. Ando, Y. Fujii, and T. Azumi,``Autoware on Board: Enabling Autonomous Vehicles with Embedded Systems,'' In Proceedings of the 9th ACM/IEEE International Conference on Cyber-Physical Systems (ICCPS2018), pp. 287-296, 2018. Link

S. Kato, E. Takeuchi, Y. Ishiguro, Y. Ninomiya, K. Takeda, and T. Hamada. ``An Open Approach to Autonomous Vehicles,'' IEEE Micro, Vol. 35, No. 6, pp. 60-69, 2015. Link
————————————————
版权声明：本文为CSDN博主「sru_alo」的原创文章，遵循CC 4.0 BY-SA版权协议，转载请附上原文出处链接及本声明。
原文链接：https://blog.csdn.net/sru_alo/article/details/88839273