Automatic Background Filtering for Cooperative Perception Using Roadside LiDAR

The vehicle-road cooperative perception needs high accuracy and real-time automatic background filtering to separate background objects from foreground objects in complex traffic scenes. Reducing the influence of foreground objects to improve accuracy, and introducing a new framework to improve real-time performance are two main challenges in automatic background filtering. This paper proposes an Automatic Background Filtering method with innovative Frame Selection and Background Matrix Extraction modules (ABF-FSBME) to address these challenges. Firstly, a new space division method with equal hitting probability is proposed to divide the 3D point cloud formed by roadside Light Detection and Ranging (LiDAR), which can reduce the influence of slight LiDAR vibrations. Secondly, the terminal-edge-cloud framework is introduced to balance delay-constrained tasks and computation-intensive tasks in automatic background filtering. Thirdly, a variance-based frame selection strategy with a sliding window mechanism is proposed to select candidate frames with fewer foreground objects. This strategy can reduce the influence of foreground objects in a coarse-grained way. Meanwhile, a new background matrix extraction method is proposed to construct the background matrix. This method can further reduce the influence of foreground objects in a fine-grained way. Finally, based on the extracted background matrix from a cloud server, the edge server can filter the raw frame in real-time. The experimental results show that the proposed ABF-FSBME method has better accuracy than other methods in error rate and integrity rate. Besides, the proposed ABF-FSBME can complete fame filtering within 10ms, and has almost no network delay, so it can satisfy the real-time requirement.

I. Introduction

With the rapid development of autonomous driving technology, connected and autonomous vehicles (CAV) is proposed and considered as one of the mainstream future directions. As is known to all, there is a defect in autonomous vehicle perception systems that only depend on the onboard sensors [1]. In the vision, CAV is easy to be misguided because it does not perceive two pedestrians blocked by the bus shown in Fig. 1. For this purpose, some researchers deploy roadside sensors to enhance the perception of CAV [2], [3]. The CAV and roadside sensors can interchange views of each other through vehicular network [4], [5] or 5/6G network [6], [7]. Cooperation with on-board sensors and roadside sensors could overcome the perceived defects of autonomous vehicles. In this way, the roadside sensors can extend the sensing range of autonomous vehicles by cooperation and improve road safety [8]. However, compared with other sensors, LiDAR has some advantages, for example, a) non-invasiveness, the deployment method of LiDAR is similar to that of cameras, without the need for digging up the roads; b) high-precision, LiDAR can offer centimeter-level positioning resolution with 10Hz~20Hz frequency, which facilitates to perceive and locate the traffic participants [9]; c) privacy protection, the output of LiDAR sensor has no personal information, such as human face, which can protect personal privacy [10]; d) good adaptability, LiDAR is not affected by illumination conditions and works well at night. Based on these advantages, LiDAR is the first choice for roadside traffic perception in this study. Fig. 1.

Cooperative Perception assisted by the Roadside Sensor.