Vehicular Localization using V2X and Information Fusion

Global navigation satellite system (GNSS) technology has been widely adopted in vehicular localisation and tracking, yet it comes with several well-known challenges in urban environments. Without direct lines of sight to at least 4 navigation satellites, GNSS localisation becomes intermittent or even corrupted. The typical accuracy of GNSS localisation in a built-up environment is more than 5 metres, which falls short of what is specified by the dedicated short-range communications (DSRC) technical committee of the Society of Automotive Engineers (SAE) - the positioning error for at least 68% of the reported vehicle positions through Basic Safety Messages (BSMs) must be less than 1.5 metres.

The introduction of 5G gives rise to opportunities for more accurate localisation not available in previous generations of cellular communications. While the use of 5G, or cellular vehicle-to-everything (C-V2X), in vehicular localisation is novel and has never been done commercially, with the existing hybrid test-bed, the project can incorporate both C-V2X and DSRC in the solution, making it more robust and universal.

Given that the position of a vehicle is an important piece of information transmitted in a vehicle-to-everything (V2X) communication network and that it plays a critical role in many V2X applications, this project aims to leverage on the V2X network to mitigate and overcome the GNSS challenges through 2 methodologies. One, to adopt or develop suitable state space models and filtering algorithms to fuse information from the roadside units (RSU) or 5G base stations (gNodeB), on-vehicle sensors like GNSS, inertia measurement unit (IMU), camera, magnetic field sensors, and map information . Two, to develop distributed cooperative methods to allow different vehicles to share information via vehicle-to-vehicle (V2V) communications to make the overall localisation of a network of vehicles more accurate and robust. Figure 1 illustrates this framework.

The outcomes of this project would provide innovative solutions using V2X and other sensors to enable vehicular localisation in GNSS-denied environments, and to augment GNSS when the signal is poor or corrupted.

As vehicular positioning plays a major role in many intelligent transport system (ITS) applications that involves safety, road toll charging, navigation and infotainment, the value of this project may come in the form of improving it in highly built-up environments like Singapore, and sheltered areas like tunnels and indoor parking lots, where GNSS can unreliable or unusable.