With the growing interest in embodied and spatial intelligence, accurately predicting trajectories in 3D environments has become increasingly critical. However, no datasets have been explicitly designed to study 3D trajectory prediction. To this end, we contribute a 3D motion trajectory (3DMoTraj) dataset collected from unmanned underwater vehicles (UUVs) operating in oceanic environments. Mathematically, trajectory prediction becomes significantly more complex when transitioning from 2D to 3D. To tackle this challenge, we analyze the prediction complexity of 3D trajectories and propose a new method consisting of two key components: decoupled trajectory prediction and correlated trajectory refinement. The former decouples inter-axis correlations, thereby reducing prediction complexity and generating coarse predictions. The latter refines the coarse predictions by modeling their inter-axis correlations. Extensive experiments show that our method significantly improves 3D trajectory prediction accuracy and outperforms state-of-the-art methods. Both the 3DMoTraj dataset and the method are available at https://github.com/zhouhao94/3DMoTraj.

Predicting how things move in three dimensions, like underwater vehicles or aerial drones, is becoming increasingly important in fields like robotics and autonomous navigation. However, most existing research only handles 2D movement because there aren’t any public datasets specifically built for 3D trajectory prediction. To address this, we introduce a new dataset called 3DMoTraj, which captures 3D motion patterns from underwater vehicles in complex ocean environments. Predicting 3D trajectories is mathematically much harder than 2D, so we also design a new method to make it easier: first, we simplify the problem by predicting each axis (x-axis, y-axis, and z-axis) separately, and then we refine the results by considering how those axes influence each other. Our experiments show that this method outperforms existing approaches and offers a strong baseline for future research.