The field of trajectory prediction and analysis is rapidly advancing, with a focus on improving the accuracy and consistency of predictions in complex scenarios. Researchers are exploring new methods for modeling trajectory data, including the use of multimodal retrieval frameworks and hierarchical self-supervised trajectory embedding frameworks. These approaches have shown promising results in capturing the complexity of human mobility and behavior in urban environments. Additionally, there is a growing emphasis on evaluating and improving the safety and reliability of autonomous driving systems, including the development of risk-based filtering approaches and deadlock avoidance testing techniques. Notable papers in this area include Universal Retrieval for Multimodal Trajectory Modeling, which introduces a novel multimodal retrieval framework, and STCLocker, which proposes a spatio-temporal conflict-guided deadlock avoidance testing technique. Furthermore, papers like Beyond Distance and HiT-JEPA demonstrate the ability to reveal invisible barriers in urban space and learn multi-scale urban trajectory representations. AMD and Improving Consistency in Vehicle Trajectory Prediction Through Preference Optimization also show potential in robust long-tail trajectory prediction and improving consistency in vehicle trajectory prediction.