Trajectory Prediction and Human Mobility

The field of trajectory prediction and human mobility is rapidly advancing, with a focus on developing more accurate and efficient models for predicting human behavior in various environments. Researchers are exploring new approaches that incorporate physics-informed constraints, variational mixture models, and cognitive risk integration to improve the accuracy and reliability of trajectory prediction. Additionally, there is a growing interest in using human body pose and social relations to enhance trajectory prediction, as well as integrating high-resolution human mobility data into epidemic modeling. Noteworthy papers in this area include PatchTraj, which proposes a dynamic patch-based trajectory prediction framework, and PhysVarMix, which presents a novel hybrid approach that integrates learning-based with physics-based constraints. Other notable papers include KASportsFormer, which introduces a kinematic anatomy-informed feature representation for 3D human pose estimation, and Social-Pose, which proposes an attention-based pose encoder for predicting human trajectories.

Sources

PatchTraj: Dynamic Patch Representation Learning for Time-Frequency Trajectory Prediction

PhysVarMix: Physics-Informed Variational Mixture Model for Multi-Modal Trajectory Prediction

RESCUE: Crowd Evacuation Simulation via Controlling SDM-United Characters

KASportsFormer: Kinematic Anatomy Enhanced Transformer for 3D Human Pose Estimation on Short Sports Scene Video

Free Energy-Inspired Cognitive Risk Integration for AV Navigation in Pedestrian-Rich Environments

Generative Active Learning for Long-tail Trajectory Prediction via Controllable Diffusion Model

Social-Pose: Enhancing Trajectory Prediction with Human Body Pose

Human Mobility in Epidemic Modeling

PAF-Net: Phase-Aligned Frequency Decoupling Network for Multi-Process Manufacturing Quality Prediction

Built with on top of