Pedestrian trajectory prediction is an important topic in autonomous vehicle (AV) research. Accurate predictions of future pedestrian locations and movements are essential for AVs to safely and efficiently navigate in dense urban environments. The recently developed Transformer (TF) network has shown superior results for applications such as natural language processing and machine translation, but limited work has been done to apply TFs for the task of pedestrian trajectory sequence prediction. In this work, we propose three novel
angle-regulated transformer models that account for trajectory characteristics such as the shape and smoothness of the trajectories. We present experimental results on real-world trajectory datasets and show that our proposed angle-regulated transformer network outperforms prior work such as recurrent neural networks and the naive TF without angle regularization. Furthermore, our method can achieve accurate prediction, even in challenging cases such as curved trajectories or sharp turns.