Vehicle Collision Avoidance in NLoS Scenarios Using Machine Learning-Assisted Positioning and MQTT

Yumei He () and Yuqin Guan ()

Abstract

In recent years, with the advancement of Auto-driving and other emerging automotive industries technology, one challenge that Road safety concerns road users, particularly the issue of collision avoidance, has become a pressing and pivotal topic. Presently, numerous devices can achieve collision avoidance in line-of-sight scenarios through sophisticated systems, Radar, Lidar, and Vision-based technologies. However, in non-line-of-sight (NLOS) scenarios, such as intersections, there is currently a lack of effective collision avoidance mechanisms. Fortunately, the integration of RTK GNSS systems with a dead reckoning solution enables high-precision positioning. Additionally, the existing 5G network provides millisecond latency-level communication between road users. Therefore, this master's project aims to explore an integrated approach incorporating the aforementioned technologies to establish an effective collision avoidance mechanism for an e-scooter and a robot vehicle in NLOS situations (e.g., intersections) and ultimately conduct a comprehensive evaluation.