关键词: T-Transformer, 图卷积网络, 锚点控制, 行人轨迹预测

Abstract: In the field of autonomous navigation, predicting pedestrian trajectories accurately is crucial for ensuring safe travel and autonomous driving. Pedestrian trajectories are highly complex, dynamic, and influenced by their surroundings, necessitating effective modeling of their temporal and spatial interactions. To address this, a model combining Transformer and Multi-Relation Graph Convolutional Networks (GCN) is proposed. It comprises an Interaction Capture Module, Anchor Control Module, and Trajectory Refinement Module. The Interaction Capture Module extracts motion features using T-Transformer and GCN, while the Anchor Control Module reduces errors by inferring intermediate destinations. Finally, the Trajectory Refinement Module enhances predictions. Experimental results on ETH and UCY datasets show superior performance in Average and Final Displacement Errors compared to mainstream models. This model's accuracy minimizes unnecessary trajectory changes and collision risks, offering a promising solution for pedestrian trajectory prediction applications.

Key words: T-Transformer , Graph Convolutional Networks , anchor control , pedestrian trajectory prediction