恶化的感知能力：用于模拟不利天气条件的自动驾驶感知能力的实时下降,arXiv - CS - Robotics

当前位置： X-MOL 学术 › arXiv.cs.RO › 论文详情

Our official English website, www.x-mol.net, welcomes your feedback! (Note: you will need to create a separate account there.)

恶化的感知能力：用于模拟不利天气条件的自动驾驶感知能力的实时下降
arXiv - CS - Robotics Pub Date : 2021-03-03 , DOI: arxiv-2103.02760
Ivan Fursa, Elias Fandi, Valentina Musat, Jacob Culley, Enric Gil, Louise Bilous, Isaac Vander Sluis, Alexander Rast, Andrew Bradley

自动驾驶汽车严重依赖于其感知子系统来观察其运行的环境。不幸的是，变化的天气条件的影响对物体检测算法提出了重大挑战，因此必须在可能遇到的所有条件下对车辆进行广泛的测试。但是，不可预测的天气可能会使在不利条件下进行真实世界的测试成为一项昂贵且耗时的工作，需要使用专业机构，并对敏感电子设备进行防风雨。仿真为现实世界的测试提供了一种替代方法，一些研究在功能强大的计算硬件上开发了越来越逼真的现实世界的可视化表示。鉴于自动驾驶车辆管道中的后续子系统并未意识到模拟的视觉真实性，当在感知下游开发模块时，外观几乎没有影响-而是感知系统在主要天气条件下的性能至关重要。这项研究探索了在自动驾驶汽车中使用简单，轻巧的图像增强系统的潜力-不仅关注视觉准确性，而是关注对感知系统性能的影响。通过最少的调整，本研究中开发的原型系统可以复制水滴在相机镜头上以及褪色的光照条件下的效果。该系统使用非常适合携带在车辆中的计算硬件，引入了小于8 ms的等待时间-使其非常适合可以在模拟实验以及现实世界中的增强现实测试期间运行的实时实施。

"点击查看英文标题和摘要"

Worsening Perception: Real-time Degradation of Autonomous Vehicle Perception Performance for Simulation of Adverse Weather Conditions

Autonomous vehicles rely heavily upon their perception subsystems to see the environment in which they operate. Unfortunately, the effect of varying weather conditions presents a significant challenge to object detection algorithms, and thus it is imperative to test the vehicle extensively in all conditions which it may experience. However, unpredictable weather can make real-world testing in adverse conditions an expensive and time consuming task requiring access to specialist facilities, and weatherproofing of sensitive electronics. Simulation provides an alternative to real world testing, with some studies developing increasingly visually realistic representations of the real world on powerful compute hardware. Given that subsequent subsystems in the autonomous vehicle pipeline are unaware of the visual realism of the simulation, when developing modules downstream of perception the appearance is of little consequence - rather it is how the perception system performs in the prevailing weather condition that is important. This study explores the potential of using a simple, lightweight image augmentation system in an autonomous racing vehicle - focusing not on visual accuracy, but rather the effect upon perception system performance. With minimal adjustment, the prototype system developed in this study can replicate the effects of both water droplets on the camera lens, and fading light conditions. The system introduces a latency of less than 8 ms using compute hardware that is well suited to being carried in the vehicle - rendering it ideally suited to real-time implementation that can be run during experiments in simulation, and augmented reality testing in the real world.

更新日期：2021-03-05

点击分享查看原文

点击收藏

阅读更多本刊最新论文

全部期刊列表>>