Leading Autonomous Transportation Toward Winter Conditions

自动驾驶汽车正在进入加拿大，为它们安全融入社会做准备的程度必须与它们预计造成的破坏程度相当。近年来，智能交通系统的进步取得了巨大的进步。这些技术飞跃是由更好的传感器和更强的计算能力推动的。这种颠覆性的创新为更多加拿大公司打开了进入新运输市场的大门，但同时也迫使政策制定者进入未知领域。尽管许多公司声称已经开发出具有高度自主性的车辆，但人们仍然怀疑这种系统在困难的气象条件下的安全性。 这项研究计划解决与自动移动系统相关的挑战，如自动驾驶汽车，与现实世界环境的复杂性有关。考虑到复杂环境的不确定性，为导航算法开辟了新的视角。在未来五年内，将特别关注与严寒冬季条件有关的具有挑战性的环境中的航行，这是一个没有得到充分考虑的问题。这项研究计划的新颖性在于将使用各种配备激光(光探测和测距，或激光雷达)的自主移动系统在野外进行广泛的实验。本地化和地图绘制方面的突破预计将推动自主移动系统在哪里以及何时能够有效和安全地使用的界限。该计划进一步分为四个具体目标：(1)将稳健优化理论融入传统的基于激光雷达的配准算法；(2)改进基于激光雷达的配准算法的不确定性估计；(3)设计更好的算法，以在复杂、动态和非结构化的环境中维护3D地图；以及(4)实现车辆必须在多个季节和困难的气象条件下自主驾驶的长期实验。 考虑到目前围绕自动驾驶汽车的兴奋情绪，该项目背后的研究具有非常重要的意义。一方面，该计划的成果将通过为复杂3D环境中的自主导航提供创新的解决方案来促进移动机器人的研究。另一方面，在向新交通解决方案过渡的关键时期，他们将通过提供关于冬季条件下自主移动系统安全的投入，为行业和政治利益攸关方提供支持。这项研究的结果还将对汽车以外的不同行业产生连锁反应，如采矿、林业和农业，这些行业在恶劣条件下需要本地化车辆，以及娱乐业，它们依赖准确的定位来实现特殊效果。

Autonomous cars are coming to Canada and preparing for their safe integration into society must be on a par with the level of disruption they are expected to cause. In recent years, advancements in intelligent transportation systems have progressed at a tremendous speed. Those technological leaps were pushed forward by better sensors and increased computation power. This disruptive innovation opens the door to a new transportation market accessible to more Canadian companies but, at the same time, forces policymakers into unknown territories. Although many companies claim to have developed vehicles with high levels of autonomy, doubts remain regarding the safety of such systems in difficult meteorological conditions. This research program addresses challenges related to autonomous mobile systems, such as autonomous cars, in relation to the complexity of real-world environments. Considering the uncertainty aspects of complex environments opens new perspectives on navigation algorithms. During the next five years, special attention will be paid to navigation in challenging environments related to harsh winter conditions, a problem which has not received sufficient consideration. The novelty of this research program lies in the extensive experiments to be performed in the field using various autonomous mobile systems equipped with a laser (light detection and ranging, or lidar). Breakthroughs in localization and mapping are expected to push the boundaries of where and when autonomous mobile systems can be used effectively and safely. This program is further divided into four specific objectives: (1) to integrate robust optimization theory into traditional lidar-based registration algorithms; (2) to improve the uncertainty estimation for lidar-based registration algorithms; (3) to design better algorithms to maintain 3D maps in the context of complex, dynamic, and unstructured environments; and (4) to realize a long-term experiment wherein a vehicle must drive autonomously through multiple seasons and difficult meteorological conditions. The research behind this program is highly significant considering the current excitement around autonomous vehicles. On the one hand, outcomes of the program will foster research in mobile robotics by providing innovative solutions for autonomous navigation in complex 3D environments. On the other hand, they will support both industrial and political stakeholders during the critical transition period toward new transportation solutions by giving inputs about the safety of autonomous mobile systems in winter conditions. The results of this research will also have a ripple effect on different industry sectors beyond automotive, such as mining, forestry, and agriculture, which required localized vehicles in harsh conditions, and the entertainment industry, which relies on accurate localization for special effects.