Human Motion Forecasting
Human motion forecasting aims to predict future human movements based on past observations, a crucial task for applications like robotics, virtual reality, and autonomous driving. Current research heavily utilizes graph convolutional networks and transformers, often incorporating contextual information such as object interactions, gaze direction, and even social dynamics between multiple individuals to improve prediction accuracy and realism. This field is advancing rapidly, driven by the development of larger, more diverse datasets and sophisticated models that address challenges like handling uncertainty and incorporating scene context for more robust and realistic predictions. The resulting improvements have significant implications for safer and more intuitive human-robot interaction and enhanced virtual and augmented reality experiences.
Papers
GaitForeMer: Self-Supervised Pre-Training of Transformers via Human Motion Forecasting for Few-Shot Gait Impairment Severity Estimation
Mark Endo, Kathleen L. Poston, Edith V. Sullivan, Li Fei-Fei, Kilian M. Pohl, Ehsan Adeli
TENET: Transformer Encoding Network for Effective Temporal Flow on Motion Prediction
Yuting Wang, Hangning Zhou, Zhigang Zhang, Chen Feng, Huadong Lin, Chaofei Gao, Yizhi Tang, Zhenting Zhao, Shiyu Zhang, Jie Guo, Xuefeng Wang, Ziyao Xu, Chi Zhang