The field of humanoid robotics and motion reconstruction is rapidly advancing, with a focus on developing more realistic and physically plausible models. Researchers are exploring new approaches to learn humanoid control policies from vision, enabling more accurate and robust motion reconstruction. Another area of focus is on improving the stability and balance of humanoid robots, particularly in multi-contact teleoperation scenarios. Additionally, there is a growing interest in developing wearable devices and sensor systems to track human motion and interaction, with applications in areas such as virtual reality, prosthetics, and ergonomic monitoring. Noteworthy papers in this area include PhysHMR, which presents a unified framework for learning visual-to-action policies for humanoid control, and Wrist2Finger, which introduces a novel wearable system for reconstructing 3D hand pose and estimating per-finger forces.
Advancements in Humanoid Robotics and Motion Reconstruction
Sources
PhysHMR: Learning Humanoid Control Policies from Vision for Physically Plausible Human Motion Reconstruction
Ego-Exo 3D Hand Tracking in the Wild with a Mobile Multi-Camera Rig
Estimation of Resistance Training RPE using Inertial Sensors and Electromyography
Wrist2Finger: Sensing Fingertip Force for Force-Aware Hand Interaction with a Ring-Watch Wearable
WAFFLE: A Wearable Approach to Bite Timing Estimation in Robot-Assisted Feeding
Stability-Aware Retargeting for Humanoid Multi-Contact Teleoperation
VBM-NET: Visual Base Pose Learning for Mobile Manipulation using Equivariant TransporterNet and GNNs
Physics-Inspired All-Pair Interaction Learning for 3D Dynamics Modeling
Learning a Shape-adaptive Assist-as-needed Rehabilitation Policy from Therapist-informed Input
ResMimic: From General Motion Tracking to Humanoid Whole-body Loco-Manipulation via Residual Learning
DPL: Depth-only Perceptive Humanoid Locomotion via Realistic Depth Synthesis and Cross-Attention Terrain Reconstruction
WristWorld: Generating Wrist-Views via 4D World Models for Robotic Manipulation