A Self-Localization Method for Urban Environments using Vehicle-Body-Embedded Off-the-Shelf Sensors

摘要

The Global Navigation Satellite System (GNSS) is by far the most used system for vehicle self-localization. However, it often fails in urban environments because of signal outages and other interference. To overcome this issue, roof-mounted LiDARs are often used by autonomous driving system developers. From a design point of view, such sensors are not appealing to the public users. Body-embedded sensors do not provide enough information because of placement and size constraints. In this paper, we show how to overcome all these constraints and achieve highly acceptable accuracy allowing reliable autonomous behavior. The proposed method's validity was demonstrated by an autonomous vehicle run in a typical urban environment.