LiDAR sensors have the advantage of being able to generate high-resolution imaging quickly during both day and night; however, their performance is severely limited in adverse weather conditions such as snow, rain, and dense fog. Consequently, many researchers are actively working to overcome these limitations by applying sensor fusion with radar and optical cameras to LiDAR. While studies on the denoising of point clouds acquired by LiDAR in adverse weather have been conducted recently, the results are still insufficient for application to autonomous vehicles because of speed and accuracy performance limitations. Therefore, we propose a new intensity-based filter that differs from the existing distance-based filter, which limits the speed. The proposed method showed overwhelming performance advantages in terms of both speed and accuracy by removing only snow particles while leaving important environmental features. The intensity criteria for snow removal were derived based on an analysis of the properties of laser light and snow particles.

Although Light Detection and Ranging (LiDAR) is a sensor type for autonomous vehicles, it is recognized as an essential tool in various fields, such as drones, Unmanned Surface Vehicles (USVs), Unmanned Ground Vehicles (UGVs), and surveillance facilities. This is because LiDAR has advantages including a high resolution, 360-degree sensing capabilities, and consistent performance during the day and at night. However, LiDAR has the fatal disadvantage of performing poorly in snow, rain, and fog. Therefore, studies are attempting to remove noise points caused by snow, rain, and fog to ensure that LiDAR performs well even under extreme weather conditions. These studies began in approximately 2020, coinciding with accelerated research on autonomous driving, and they are still being actively pursued today. In particular, with the development of artificial intelligence technology in addition to the existing conventional methods, various approaches are being developed, but papers that thoroughly analyze and organize this research have not yet been published. Accordingly, in this study, we aim to comprehensively review the studies that have attempted to remove LiDAR sensor noise under extreme weather conditions through various methods. This review discusses the latest LiDAR denoising algorithms, with a particular focus on techniques implemented under snowy conditions. These algorithms can be categorized into five types: distance-based, intensity-based, fusion-based, learning-based, and other methods. In addition, this paper covers extreme weather datasets and related perception studies.

Studies on autonomous driving have started to focus on snowy environments, and studies to acquire data and remove noise and pixels caused by snowfall in such environments are in progress. However, research to determine the necessary weather information for the control of unmanned platforms by sensing the degree of snowfall in real time has not yet been conducted. Therefore, in this study, we attempted to determine snowfall information for autonomous driving control in snowy weather conditions. To this end, snowfall data were acquired by LiDAR sensors in various snowy areas in South Korea, Sweden, and Denmark. Snow, which was extracted using a snow removal filter (the LIOR filter that we previously developed), was newly classified and defined based on the extracted number of snow particles, the actual snowfall total, and the weather forecast at the time. Finally, we developed an algorithm that extracts only snow in real time and then provides snowfall information to an autonomous driving system. This algorithm is expected to have a similar effect to that of actual controllers in promoting driving safety in real-time weather conditions.

Over the past decades, significant efforts have been dedicated to the development of ideal actuators for robotic systems, focusing on both mechanisms and design aspects [1], and they are implemented as a form of electric motors and hydraulic/pneumatic actuators. In the case of electric motors, they have emerged as a widely adopted choice for actuating robotic systems due to their compactness, ease of implementation, excellent linearity, and high-control bandwidth. Consequently, they remain one of the favorable compromise solutions for general robotic applications, offering exceptional dynamic output and precise control [2], [3]. In addition, electric motors exhibit a higher energy per weight or volume density than that of the human musculoskeletal system [4], [5], making them apparently advantageous for robotic applications.