Research on Multimodal Fusion Target Motion Trajectory Prediction Algorithm Based on Deep Reinforcement Learning

Trajectory prediction for moving targets is a key technology in fields such as intelligent manufacturing, autonomous driving, and video surveillance. This paper proposes a Deep Multimodal Fusion Trajectory Prediction (DMFTP) algorithm based on deep reinforcement learning for multimodal fusion target motion trajectory prediction. The algorithm establishes a multimodal trajectory prediction model with an attention mechanism to reduce the information loss and difficulties in modeling in multimodal fusion target motion trajectory prediction. Tested on multiple datasets, the results show that the algorithm has certain advantages over existing technologies in terms of accuracy and robustness. The research in this paper aims to provide a feasible solution for target motion trajectory prediction in complex environments.