Collaborative localization of vehicle formations based on ranges and bearings

We examine the problem of jointly determining the positions of multiple underwater vehicles based on a set of pairwise range and bearing measurements taken over time. This extends prior work on the so-called (static) collaborative localization paradigm where a hybrid approach was proposed for seamless instantaneous fusion (i.e., no time dependence) of range and bearing measurements. To incorporate time we add to the original convexified least-squares cost function a regularizing term that penalizes deviations between predicted and computed vehicle positions at a given instant. The method operates progressively over time, with past estimates used for prediction at the current instant assuming a very simple quasilinear motion model. The method is amenable to parallelization, with simple gradient-like updates. Numerical results demonstrate promising accuracy gains (reduction on the order of 10 % in terms of root-mean-square positioning error) in simulations inspired by an underwater geoacoustic surveying application.

Access paper here

Bibtex:

@inproceedings{Ferreira_2016,
    author = "Ferreira, Beatriz Quintino and Gomes, Joao and Soares, Claudia and Costeira, Joao P.",
    title = "Collaborative localization of vehicle formations based on ranges and bearings",
    url = "http://dx.doi.org/10.1109/ucomms.2016.7583426",
    DOI = "10.1109/ucomms.2016.7583426",
    booktitle = "2016 IEEE Third Underwater Communications and Networking Conference (UComms)",
    publisher = "IEEE",
    year = "2016",
    month = "Aug",
    pages = "1-5",
    abstract = "We examine the problem of jointly determining the positions of multiple underwater vehicles based on a set of pairwise range and bearing measurements taken over time. This extends prior work on the so-called (static) collaborative localization paradigm where a hybrid approach was proposed for seamless instantaneous fusion (i.e., no time dependence) of range and bearing measurements. To incorporate time we add to the original convexified least-squares cost function a regularizing term that penalizes deviations between predicted and computed vehicle positions at a given instant. The method operates progressively over time, with past estimates used for prediction at the current instant assuming a very simple quasilinear motion model. The method is amenable to parallelization, with simple gradient-like updates. Numerical results demonstrate promising accuracy gains (reduction on the order of 10 \\% in terms of root-mean-square positioning error) in simulations inspired by an underwater geoacoustic surveying application.",
    keywords = "distance measurement;least squares approximations;mean square error methods;underwater acoustic communication;underwater vehicles;bearing measurements;collaborative localization paradigm;computed vehicle positions;convexified least-squares cost function;multiple underwater vehicle positions;pairwise range measurements;predicted vehicle positions;quasilinear motion;root-mean-square positioning error;seamless instantaneous fusion;simple gradient-like updates;underwater geoacoustic surveying application;vehicle formations;Acoustics;Collaboration;Cost function;Position measurement;Time measurement;Trajectory;Vehicles"
}