Imitation Learning of Hierarchical Driving Model: From Continuous Intention to Continuous Trajectory
Imitation Learning of Hierarchical Driving Model: From Continuous Intention to Continuous Trajectory
复制标题
分层驾驶模型的模仿学习：从连续意图到连续轨迹
DOI：
10.1109/lra.2021.3061336
复制
发表时间：
2020-10
影响因子：
5.2
通讯作者：
Xiong Rong
中科院分区：
文献类型：
--
作者：
Wang Yunkai;Zhang Dongkun;Wang Jingke;Chen Zexi;Li Yuehua;Wang Yue;Xiong Rong
如何赋予系统处理环境、意图和动态耦合复杂性的学习能力，是缩小机器与人类水平驾驶之间差距的挑战之一。在这封信中，我们提出了一个具有连续意图和连续动力学的显式模型的分层驾驶模型，该模型解耦了人类驾驶数据中观察到行动推理的复杂性。具体来说，连续意图模块利用感知来生成包含障碍和意图的潜在地图。然后，将势映射作为条件，结合当前动态，通过连续函数逼近网络生成连续轨迹作为输出，该网络的导数可用于监督，无需附加参数。最后，通过数据集和模拟实验验证了该方法的有效性，结果表明该方法具有更高的位移和速度预测精度，生成的轨迹更平滑。我们的方法也被部署在具有循环延迟的真实车辆上，验证了其有效性。据我们所知，这是第一个使用连续函数逼近网络产生驱动轨迹的工作。我们的代码可在https://github.com/ZJU-Robotics-Lab/CICT上获得。
One of the challenges to reduce the gap between the machine and the human level driving is how to endow the system with the learning capacity to deal with the coupled complexity of environments, intentions, and dynamics. In this letter, we propose a hierarchical driving model with explicit models of continuous intention and continuous dynamics, which decouples the complexity in the observation-to-action reasoning in the human driving data. Specifically, the continuous intention module takes perception to generate a potential map encoded with obstacles and intentions. Then, the potential map is regarded as a condition, together with the current dynamics, to generate a continuous trajectory as output by a continuous function approximator network, whose derivatives can be used for supervision without additional parameters. Finally, our method is validated by both datasets and stimulation, demonstrating that our method has higher prediction accuracy of displacement and velocity and generates smoother trajectories. Our method is also deployed on the real vehicle with loop latency, validating its effectiveness. To the best of our knowledge, this is the first work to produce the driving trajectory using a continuous function approximator network. Our code is available at https://github.com/ZJU-Robotics-Lab/CICT.
登录
查看更多内容
DeepGoal：从人类控制演示中学习驾驶意图
影响因子：
4.3
作者：
Ma Huifang;Wang Yue;Xiong Rong;Kodagoda Sarath;Tang Li
通讯作者：
Tang Li
DOI：
10.1609/aaai.v33i01.33016120
发表时间：
2018-11
期刊：
ArXiv
影响因子：
--
作者：
Yuexin Ma;Xinge Zhu;Sibo Zhang;Ruigang Yang;Wenping Wang;Dinesh Manocha
通讯作者：
Yuexin Ma;Xinge Zhu;Sibo Zhang;Ruigang Yang;Wenping Wang;Dinesh Manocha
预见未来：用单目相机预测自我车辆轨迹的语义分割网络
影响因子：
5.2
作者：
Sun, Yuxiang;Zuo, Weixun;Liu, Ming
通讯作者：
Liu, Ming
影响因子：
25
作者：
Brady Zhou;Philipp Krähenbühl;V. Koltun
通讯作者：
Brady Zhou;Philipp Krähenbühl;V. Koltun
DOI：
--
发表时间：
2018
期刊：
--
影响因子：
--
作者：
Yuexin Ma;Xinge Zhu;Sibo Zhang;R. Yang;Wenping Wang;Dinesh Manocha
通讯作者：
Yuexin Ma;Xinge Zhu;Sibo Zhang;R. Yang;Wenping Wang;Dinesh Manocha