DeepGoal: Learning to drive with driving intention from human control demonstration
DeepGoal: Learning to drive with driving intention from human control demonstration
复制标题
DeepGoal：从人类控制演示中学习驾驶意图
DOI：
10.1016/j.robot.2020.103477
复制
发表时间：
2019-11
影响因子：
4.3
通讯作者：
Tang Li
中科院分区：
文献类型：
--
作者：
Ma Huifang;Wang Yue;Xiong Rong;Kodagoda Sarath;Tang Li
最近对汽车驾驶的研究开发了一种高效的端到端学习模式，直接将视觉输入映射到控制命令。然而，它在单个网络中对不同的驱动变量进行建模，这增加了学习复杂性，并且不太适合模块集成。在本文中，我们重新研究了人类的驾驶风格，并提出了学习一个中间驾驶意向区域来缓解端到端方法的困难。在公共路径规划器中，意图区域在图像和指向目标的方向上都遵循道路结构，它只处理视觉变化，并且在没有传统的精确定位的情况下计算出要去哪里。然后将学习到的视觉意图投影到车辆局部坐标上，并与可靠的障碍物感知融合，生成广泛用于运动规划的导航记分图。该系统的核心是一个弱监督的cGAN-LSTM模型，该模型被训练成从人类演示中学习驾驶意图。对抗性损失从具有一条本地规划路线的有限演示数据中学习，并在测试过程中允许对具有不同路线的多模式行为进行推理。使用真实世界的数据集进行了全面的实验。结果表明，该模型能产生与人的演示更一致的运动指令，对环境变化具有更好的可靠性和鲁棒性。我们的代码可以在https://github.com/HuifangZJU/visual-navigation.上找到
Recent research on automotive driving has developed an efficient end-to-end learning mode that directly maps visual input to control commands. However, it models distinct driving variations in a single network, which increases learning complexity and is less adaptive for modular integration. In this paper, we re-investigate human’s driving style and propose to learn an intermediate driving intention region to relax the difficulties in end-to-end approach. The intention region follows both road structure in image and direction towards goal in public route planner, which addresses visual variations only and figures out where to go without conventional precise localization. Then the learned visual intention is projected on vehicle local coordinate and fused with reliable obstacle perception to render a navigation score map that is widely used for motion planning. The core of the proposed system is a weakly-supervised cGAN-LSTM model trained to learn driving intention from human demonstration. The adversarial loss learns from limited demonstration data with one local planned route and enables reasoning of multi-modal behaviors with diverse routes while testing. Comprehensive experiments are conducted with real-world datasets. Results indicate the proposed paradigm can produce more consistent motion commands with human demonstration and shows better reliability and robustness to environment change. Our code is available at https://github.com/HuifangZJU/visual-navigation.
登录
查看更多内容
影响因子：
6
作者：
Qianhui Luo;Huifang Ma;Li Tang;Yue Wang;R. Xiong
通讯作者：
Qianhui Luo;Huifang Ma;Li Tang;Yue Wang;R. Xiong
DOI：
10.1109/icccnt56998.2023.10306417
发表时间：
2022-02
期刊：
2023 14th International Conference on Computing Communication and Networking Technologies (ICCCNT)
影响因子：
--
作者：
Gilad Cohen;Raja Giryes
通讯作者：
Gilad Cohen;Raja Giryes
DOI：
--
发表时间：
2016-11
期刊：
ArXiv
影响因子：
--
作者：
Ari Seff;Jianxiong Xiao
通讯作者：
Ari Seff;Jianxiong Xiao
DOI：
--
发表时间：
2016-04
期刊：
ArXiv
影响因子：
--
作者：
Mariusz Bojarski;D. Testa;Daniel Dworakowski;Bernhard Firner;B. Flepp;Prasoon Goyal;L. Jackel;
通讯作者：
Mariusz Bojarski;D. Testa;Daniel Dworakowski;Bernhard Firner;B. Flepp;Prasoon Goyal;L. Jackel;
DOI：
--
发表时间：
2017-10
期刊：
--
影响因子：
--
作者：
Wei Gao-;David Hsu;Wee Sun Lee;Shengmei Shen;K. Subramanian
通讯作者：
Wei Gao-;David Hsu;Wee Sun Lee;Shengmei Shen;K. Subramanian