WAVES - Waveform Analysis for Vehicles in Extreme Scenarios

激光雷达是一种传感器，它利用光来测量到反射物体的距离，并生成周围环境的3D表示。它们是通常安装在自动驾驶汽车上的关键传感器，用于检测障碍物并准确定位道路上的车辆。目前的趋势是使用越来越高清晰度的传感器，从而产生大量的数据。反过来，这一趋势又推动了整车成本的上升。此外，这些高清晰度激光雷达通常依赖于旋转电机来扫描周围环境，这使得它们对由于振动而导致的过早故障很敏感。这一弱点是对自动驾驶汽车长期安全性的担忧。另一种方法是使用没有移动部件的传感器，称为固态激光雷达。LeddarTech和拉瓦尔大学之间的这项研究合作旨在使用这些固态激光雷达提高北欧条件下自动驾驶汽车的感知能力。该计划的第一个目标是使用机器学习算法来更好地检测和识别道路上的移动障碍物。第二个目标是将连续强度测量集成到概率映射算法中。这两个目标都将在冬季条件下得到验证，以确保低能见度条件下的安全。这项研究计划将通过简化自动驾驶汽车技术在加拿大驾驶条件下的整合来影响交通系统。

Lidars are sensors using light to measure distances to reflecting objects and to generate a 3D representation of the surroundings. They are critical sensors commonly installed on autonomous cars to detect obstacles and accurately locate vehicles on roads. The current trend is to use increasingly higher definition sensors, thus producing a large amount of data. In turn, this trend drives the cost upward for the entire vehicle. Moreover, these high-definition lidars generally rely on a spinning motor to scan the surroundings, which make them sensitive to premature failure due to vibrations. This weakness is a concern for long-term safety of autonomous vehicles. A different approach is to use sensors without moving parts named solid-state lidars. This research collaboration between LeddarTech and Laval University aims at improving perception capabilities of autonomous vehicles in Nordic conditions using these solid-state lidars. The first objective of the program is to use machine learning algorithms to better detect and identify moving obstacles on the road. The second objective is to integrate continuous intensity measurements into probabilistic mapping algorithms. Both of these objectives will be validated against winter conditions to ensure safety in low- visibility conditions. This research program will impact the transportation system by easing the integration of autonomous vehicle technology in Canadian driving conditions.