Propagating and measuring anchor uncertainty in space-time prisms on road networks

Abstract

Space-time prisms capture all possible spatio-temporal locations of a moving object between sample points given speed limit constraints on its movement. These sample points are usually considered to be perfect measurements. In this paper we restrict ourselves to a road network and extend the notion of sample points to sample regions, which are bounded, sometimes disconnected, subsets of space-time wherein each point is a possible location, with its respective probability, where a moving object could have originated from or arrived in. This model allows us to model measurement errors, multiple possible simultaneous locations and even flexibility of a moving object. We develop an algorithm that computes the envelope of all space-time prisms that have an anchor in these sample regions and we developed an algorithm that computes for any spatio-temporal point the probability with which a space-time prism, with anchors in these sample regions, contains that point. We implemented these algorithms in Mathematica to visualise all these newly-introduced concepts.