The field of human motion tracking and analysis is moving towards more accurate and robust methods for estimating human pose, gait, and terrain geometry. Recent developments have focused on leveraging novel sensor modalities, such as wearable soft sensors and radar data, to improve the accuracy and comfort of motion tracking systems. Additionally, there is a growing interest in developing frameworks that can efficiently adapt to new users and environments, as well as quantify uncertainty in pose estimates. Noteworthy papers in this area include:

• A model-agnostic meta-learning framework for adaptive gait phase and terrain geometry estimation, which demonstrates superior accuracy and adaptation efficiency.
• A real-time human motion tracking method that combines IMU and barometric data to estimate human pose and global translation on uneven terrain, outperforming existing methods that use IMUs only.
• A probabilistic encoder-decoder architecture for radar-based human pose estimation, which achieves state-of-the-art performance and provides uncertainty quantification for pose estimates.
• A physics-informed trajectory compression framework that supports trajectories in arbitrary dimensions and achieves superior compression ratios and fidelity compared to existing methods.