Dichotomy: Trajectory Planning Algorithm Based on Point Group Distribution

Abstract

The historical navigation trajectory carries rich information about the spatiotemporal patterns of a moving target, as well as its transboundary potentials. To obtain the moving track of a ship or a plane, a large amount of point-to-point data is required via the measurements of satellites or aviation equipments. On the basis of time, we can easily draw a track. However, the distribution of the point-to-point data information provided by each detection unit is irregular due to the inconsistency of the standards of the data units. Some of the regional point groups are densely distributed, while the others may be very sparse. The point density can mess up the trajectory detection and make it not ideal to observe the behavior habits of a target unit. In this study, a dichotomy trajectory planning algorithm was proposed to fix the problem mentioned above. The target area was set as a grid with the points in the grid dichotomized. In that case, the distribution of the targeted point group after dichotomization can perform as similar as possible, and an approximate trajectory line was formed. Such a method considers the change of point density. Results were tested against the performance of the clustering, the random, and the greedy methods, showing that the former had a better adaptability in dealing with different data distribution. In this paper, a novel approximate estimation algorithm based on the density change rate is proposed. Finally, we will compare the performance of our dichotomous method with clustering method, random method, and greedy algorithm through experiments and show the trajectory fitting effect through different distribution data.