A novel UAV path planning algorithm to search for floating objects on the ocean surface based on object’s trajectory prediction by regression

Search and find mission in ocean environment is a none trivial operation given the amount of random parameters associated with it. The uncertain and dynamic aspects related to ocean current movement make the trajectory prediction of drifting lost object onto sea water a very complicated task. In this work we present a novel lost target searching algorithm based on Recursive Area Clustering and target trajectory predication in ocean environment. Based on the widely known GlobCurrent v2 dataset which model the drifting of ocean surface current using satellite sensory data combined with mathematical and simulation modeling, we propose a regression algorithm based on our Recursive Area Clustering algorithm that we have developed previously to determine the strategic zones (weight centers) characterizing the high density areas extracted from drifting target history. Given those weight centers, we predict the object trajectory through refined regression. The predicted lost object trajectory is used to plan the path of UAV search mission. The model developed has a significant impact as we have tested our strategy in a scenario for searching an area covering 68517 km2, we have shown that 78% of the time, the lost object can be found within 32 km distance of the predicted trajectories limiting the significant search area to be about 5% of the whole searched area. •In this work we presented a novel lost target searching algorithm based on Recursive Area Clustering and target trajectory predication in ocean environment.•We used the widely known GlobCurrent v2 dataset which model the drifting of ocean surface current using satellite sensory data combined with mathematical and simulation modeling.•We proposed a regression algorithm based on our Recursive Area Clustering algorithm that we have developed previously to determine the strategic zones (weight centers) characterizing the high density areas extracted from drifting target history.•Given those weight centers, we predict the object trajectory through refined regression. The predicted lost object trajectory is used to plan the path of UAV search mission.•We obtained encouraging results with a significant impact as we have tested our strategy in a scenario for searching an area covering 68517 km2, we have shown that 78% of the time, the lost object can be found within 32 km distance of the predicted trajectories limiting the significant search area to be about 5% of the whole searched area.