System for Portable Body Surface Acquisition


  • Portable wireless 3D posture monitoring for back pain management
  • Portable monitoring of 4D posture (3D + time) for kinesiology
  • Injury prevention and injury risk detection in military applications, weight lifting, etc.
  • Artificial proprioception for non-rigid robots
  • Entertainment industry: avataring and on-field virtual reality gaming

Problem Addressed

Existing commercial technology based on Inertial Measurement Units (IMUs) for portable motion capture is limited to articulated rigid models (i.e. skeletons).  IMUs provide reliable 3D-orientation measurements which can be interpreted in terms of rigid segments and joint angles, given a template skeleton model.  However, the surface of a non-rigidly deforming object such as the human body cannot be trivially reconstructed from orientations.  Existing methods sometimes simplify the surface geometry of such objects into piece-wise rigid geometries, which results in low accuracy in the resulting reconstruction.  Other approaches combine orientation readings with other technologies that can provide absolute position information.  Unfortunately, the combination of IMU sensors with other modalities to capture non-rigid movement tends to make the system cumbersome and impractical, as indicated by the lack of commercial solutions along this line.


This invention presents a portable 3D imaging system that can acquire human body movement beyond rigid joint-angle measurement.  Using an ensemble of orientation sensors (IMUs), this technology can capture non-rigid deformations of the body surface.  The system is comprised of a tight-fitting textile undergarment with embedded MEMS IMU devices (each comprising a 3D accelerometer, a 3D magnetometer, and a 3D gyroscope), 3D video capturing devices for calibration, and novel computer algorithms.  This is a high performance training-based approach with high robustness and scalable accuracy. 


  • Portable wireless, markerless, cameraless 3D imaging of non-rigidly deforming human body surface
  • More accurate and robust than existing technology with scalable accuracy depending on the number of embedded sensors
  • Delivers real-time performance
  • Adapts to unique geometry and biomechanics of each individual thanks to the underlying machine learning technique employed