Prediction and retrodiction algorithms for path-constrained targets

This paper presents algorithms for prediction and retrodiction of targets whose motion is constrained by external conditions (e.g., shipping lanes, roads). The targets are moving along paths defined by waypoints and segments. Measurements are obtained by sensors (e.g., spaceborne) at low revisit rates. Existing tracking algorithms assume that the targets follow the same motion model between successive measurements, but in a low revisit rate scenario targets may change the motion model between successive measurements. The proposed prediction algorithm addresses this issue by considering the possible changes in the motion model whenever targets move to a different segment. Further, when a target approaches an intersection, it has the possibility to travel into one of the multiple segments beyond that intersection. To predict the probable locations, multiple hypotheses for segments are introduced and a probability is calculated for each segment hypothesis. When measurements become available later, segment hypothesis probabilities are updated based on a combined mode likelihood and a sequential probability ratio test (SPRT) is carried out to reject the unlikely hypotheses. Retrodiction for path-constrained targets is also considered, because in some scenarios it is desirable to find out a target's location at some previous time (e.g., at the time of an accident) given all subsequent measurements. A retrodiction algorithm is also developed for path-constrained targets so as to facilitate motion forensic analysis. Simulation results and a performance measure for path-constrained target tracking are presented to validate the proposed algorithms.