The author introduces a Recurrent Aligned Network (RAN) to address the domain shift problem in pedestrian trajectory prediction by minimizing the domain gap through domain alignment.
The author presents a dynamics-based deep learning framework integrating an asymptotically stable dynamical system into a Transformer model for pedestrian trajectory prediction, aiming to provide explainability and enforce explicit constraints on predicted trajectories.
The author introduces LG-Traj, a novel approach utilizing Large Language Models (LLMs) to enhance pedestrian trajectory prediction by incorporating motion cues from past and future trajectories. The method integrates motion patterns and social interactions for accurate trajectory forecasting.
Large Language Models (LLMs) are utilized in LG-Traj to enhance pedestrian trajectory prediction by incorporating motion cues and social interactions, demonstrating effectiveness on benchmark datasets.
Introducing a Recurrent Aligned Network for generalized pedestrian trajectory prediction to address domain shift challenges.