Cooperative Vehicle Localization
Mark Christopher Johnson and Kannan Ramchandran
National Science Foundation 0729237 and National Science Foundation 0635114
Accurate and reliable vehicle localization is a key component of numerous applications, including active vehicle safety systems, real time estimation of traffic conditions, and high occupancy tolling. In this project, we explore novel methods of localizing vehicles on roads, both using fixed infrastructure and in an infrastructure-free environment. We examine the measurement of the ranges and angles of vehicles, and the estimation of vehicle position from noisy measurements. We consider localization methods that incorporate fixed base stations and pseudolites, ground based transmitters that can augment or replace the GPS system. In particular, we study algorithms based on multi-antenna receivers that measure the angle-of-arrival from pseudolites. We characterize the error performance in a multipath environment, and show that accurate estimates can be achieved with a suitably high sampling rate. We also consider the problem of cooperative localization in the absence of fixed infrastructure. We propose an iterative scheme for collaborative localization of a set of vehicles, which does not require base stations, based on belief propagation. This is especially useful in rural areas and developing countries.
More information: http://www.eecs.berkeley.edu/~mjohnson/