View article

Data representing moving objects is rapidly getting more available, especially in the area of wildlife GPS tracking. It is a central belief that information is hidden in large data sets in the form of interesting patterns, where a pattern can be any configuration of some moving objects in a certain area and/or during a certain time period. One of the most common spatio-temporal patterns sought after is flocks. A flock is a large enough subset of objects moving along paths close to each other for a certain pre-defined time. We give a new definition that we argue is more realistic than the previous ones, and by the use of techniques from computational geometry we present fast algorithms to detect and report flocks. The algorithms are analysed both theoretically and experimentally.