Detecting movement patterns using Brownian bridges

In trajectory data a low sampling rate leads to high uncertainty in between sampling points, which needs to be taken into account in the analysis of such data. However, current algorithms for movement analysis ignore this uncertainty and assume linear movement between sample points. In this paper we develop a framework for movement analysis using the Brownian bridge movement model (BBMM), that is, a model that assumes random movement between sample points. Many movement patterns are composed from basic building blocks, like distance, speed or direction. We efficiently compute their distribution over space and time in the BBMM using parallel graphics hardware. We demonstrate our framework by computing patterns like encounter, avoidance/attraction, regular visits, and following. Our motivation to study the BBMM stems from the rapidly expanding research paradigm of movement ecology. To this end, we provide an interface to our framework in R, an environment widely used within the natural sciences for statistical computing and modeling, and present a study on the simultaneous movement of groups of wild and free-ranging primates.