| 本文已被：浏览 2019次 下载 4803次 |
码上扫一扫！ |
|
|
| 基于凸优化的无人驾驶汽车转向角安全性验证 |
|
吴慧慧1,2, 张亚楠3, 侯刚1,2, 渡边政彦4, 王洁1,2, 孔维强1,2
|
|
1.大连理工大学 软件学院, 辽宁 大连 116024;2.辽宁省泛在网络与服务软件重点实验室, 辽宁 大连 116024;3.中汽数据有限公司, 天津 300300;4.NTT DATA Automobiligence Research Center, Yokohama 222-0033, Japan
|
|
| 摘要: |
| 无人驾驶汽车系统过大的输入-输出空间（即输入和输出的所有可能组合），使得为其提供形式化保证变成一项具有挑战性的任务.提出了一种自动验证技术，通过结合凸优化和深度学习验证工具DLV来保障无人驾驶汽车的转向角安全.DLV是一个用于自动验证图像分类神经网络安全性的框架.运用故障安全轨迹规划中的凸优化技术解决预测转向角的判断问题，然后拓展DLV来实现无人驾驶汽车转向角安全性的验证.在NVIDIA的端到端无人驾驶架构上说明所提出方法的优势，该架构是许多现代无人驾驶汽车的关键组成部分.实验结果表明：对于给定的区域和操作集，如果存在对抗性错误分类（即不正确的转向决策），该技术可以成功地找到，因此可以实现安全验证（如果在所有DNN层都没有发现错误分类，在这种情况下，网络关于转向决策可以说是稳定或可靠的）或证伪（在这种情况下，这些对抗性反例可以用于后续微调网络）. |
| 关键词: 无人驾驶汽车 转向角 自动驾驶汽车 凸优化 安全性验证 |
| DOI：10.13328/j.cnki.jos.006851 |
| 分类号: |
| 基金项目:国家重点研发计划（2020YFB2009500）；中央高校基本科研业务费专项资金（DUT20TD107，DUT22ZD203）；NTTDATA智能汽车研究所 |
|
| Verification of Steering Angle Safety for Self-driving Cars Using Convex Optimization |
|
WU Hui-Hui1,2, ZHANG Ya-Nan3, HOU Gang1,2, WATANABE Masahiko4, WANG Jie1,2, KONG Wei-Qiang1,2
|
|
1.School of Software Technology, Dalian University of Technology, Dalian 116024, China;2.Key Laboratory for Ubiquitous Network and Service Software of Liaoning Province, Dalian 116024, China;3.Automotive Data of China (Tianjin) Co. Ltd., Tianjin 300300, China;4.NTT DATA Automobiligence Research Center, Yokohama 222-0033, Japan
|
| Abstract: |
| Providing formal guarantees for self-driving cars is a challenging task, since input-output space (i.e., all possible combinations of inputs and outputs) is too large to explore exhaustively. This paper presents an automated verification technique ensuring steering angle safety for self-driving cars by incorporating convex optimization and deep learning verification (DLV). DLV is an automated verification framework for safety of image classification neural networks. The DLV is extended by convex optimization technique in fail-safe trajectory planning to solve the judgement problem of predicted steering angle, and thus, to achieve verification of steering angle safety for self-driving cars. The benefits of the proposed approach are demonstrated on the NVIDIA's end-to-end self-driving architecture, which is a crucial ingredient in many modern self-driving cars. The experimental results indicate that the proposed technique can successfully find adversarial misclassifications (i.e., incorrect steering decisions) within given regions and family of manipulations if they exist. Therefore, the safety verification can be achieved (if no misclassification is found for all DNN layers, in which case the network can be said to be stable or reliable w.r.t. steering decisions) or falsification (in which case the adversarial examples can be used to fine-tune the network). |
| Key words: self-driving cars steering angle autonomous vehicle convex optimization safety verification |
