A fundamental capability of any robot is that of localizing itself in the environment. Providing this capability to autonomous industrial vehicles that operate outdoors poses significant challenges, as the environment is typically unstructured and changing over time.

The solution that is most typically used in current practice is GPS – however, GPS coverage is poor in many relevant industrial environments, such as open pit mines, below silos, and in forested areas. In case study, we work with Volvo Construction Equipment to develop a method that provides pose estimates of machinery without relying on GPS data. The aim is to provide a redundant system that can be installed alongside the commonly used RTK-GPS based positioning systems. We utilize range-measuring devices and fuse their measurements with our probabilistic map representation framework to compute pose estimates.