Dynamic base station DGPS for Cooperative Vehicle Localization Mohsen Rohani, Denis Gingras Electrical and Computer Engineering Department Laboratory on Intelligent Vehicles Université de Sherbrooke, Sherbrooke, Qc, Canada mohsen.rohani@usherbrooke.ca, denis.gingras@usherbrooke.ca Dominique Gruyer IFSTTAR - CoSys - LIVIC Versailles, France dominique.gruyer@ifsttar.fr Abstract— Cooperative approaches are becoming of great interest in automotive research. One of the most important ITS applications is cooperative localization. In this paper, the concept of a dynamic base station DGPS (DDGPS) and its application in the vehicular cooperative localization is introduced and discussed. The DDGPS is a decentralized cooperative method which aims to improve the GPS positioning by estimating and compensating the common error in GPS pseudorange measurements. It can be seen as an extension of DGPS where the base stations are not necessarily static with an exact known position. In the DDGPS method, the pseudorange corrections are estimated, based on the receiver’s belief on its positioning and its uncertainty, and then broadcasted to other GPS receivers. A new method for fusing all the received corrections from different sources is proposed and the data dependency problem is also discussed. Keywords— DGPS, dynamic base station DGPS, DDGPS, Cooperative localization, GPS. I. INTRODUCTION Navigation systems constitute an essential component of intelligent vehicles and are being used in a great variety of active or informative ADAS applications. GNSS based navigation systems allow to easily obtain information on the absolute position of the vehicle and their use are widely spread in ITS applications [1]. However, low cost GPS receiver- based navigation systems used in automotive applications suffer from low accuracy and frequent signal outages. Typically, the GPS nominal accuracy is about 15m, which is usually not sufficient for active safety and ADAS applications such as lane level positioning. One of the most common ways to improve accuracy for ego-localization is to use other embedded sources of information and to combine them with GNSS data. Those other sources can be dead reckoning sensors, such as INS and odometer, or video cameras [2,3]. This approach typically use data fusion algorithms, like Kalman filters or particle filters [4] to combine the information of those different sensors in order to obtain a better position estimate than the one obtained by the GPS receiver alone or by each of the individual sensor. A classical approach to enhance the GPS positioning accuracy is to use a differential method exploiting a fixed known position as a ground based reference, hence the name differential GPS (DGPS) [5]. In DGPS, the ground based reference station with an exactly known position, broadcasts its GPS receiver information, which allows to calculate and correct the errors of the measured pseudoranges obtained by other non-fixed GPS receivers in the vicinity. The method exploits the fact that GPS receivers, which are close to each other, are affected by the various sources of errors in a similar way. This assumption can be done because of the use of the same set of satellites in order to assess ego-localisation. To apply this approach in the real world and with static road side stations it requires to deploy a large number of reference stations in order to be able to enhance the GPS position in a given region. This approach is therefore very expensive in terms of infrastructure. In addition DGPS to operate properly always requires a communication link between the reference stations and the mobile GPS receivers. These two constraints make the DGPS approach difficult to implement and also very expensive to use for general vehicle positioning in automotive applications. However, with the recent emergence of wireless communication capabilities and VANETS, cooperative positioning is becoming an attractive alternative for improving positioning performance [6-10]. The main goal of cooperative localization is to exploit different sources of information coming from not only an ego-vehicle but different vehicles within a short range area, in order to enhance positioning system efficiency while keeping the computing and infrastructure costs at a reasonable level. In this paper we aim to improve the GPS vehicle position estimates by exploiting position information from other vehicles or mobile objects (pedestrian, bicycle etc.). The basic idea is to extend the DGPS method by using mobile reference stations instead of fixed one, thus generating pseudo-range corrections by nearby vehicles and broadcasting them to be used by nearby vehicles. This idea brings challenges as the mobile reference stations do not have a precisely known position, and therefore, the pseudo-range corrections generated by them also suffer from significant uncertainties. In the next section, the concept of dynamic base station DGPS (DDGPS) will be introduced and discussed. The proposed method can be seen as an extension of the DGPS method, where the base stations don’t need to be static and