BaroPoser: Real-time Human Motion Tracking from IMUs and Barometers in Everyday Devices
summary
BaroPoser fuses IMU and barometer readings from two everyday devices (smartwatch on the wrist, smartphone in the side pocket) to output full-body pose and global translation in real time on non-flat terrain. It builds a side-rooted local thigh frame to decouple local pose from global orientation, computes device height from barometric pressure (sensitive to 5–10 cm height changes), predicts horizontal velocity with a neural net and vertical motion from barometer-derived height, and outperforms IMU-only SOTA on benchmarks and a newly collected dataset.
AI-written summary · about
chapters (9)
- 0:00Introduction to Bioposer
- 1:59Background on inertial motion capture
- 3:43Method overview and modules
- 4:25Data pre-processing details
- 5:06Local pose estimation approach
- 5:51Global translation estimation
- 6:12Experimental results and datasets
- 7:35Ablation studies and conclusions
- 7:56Q&A session
related talks
06mIMUTube: Automatic extraction of virtual on-body accelerometry from video for human activity ...
10mProactive Sensing for Improving Hand Pose Estimation
22mRunning, walking, sitting or biking? - Motion prediction with acceleration and rotation
06mRobust Inertial Motion Tracking through Deep Sensor Fusion across Smart Earbuds and Smartphone
06mWR-Hand: Wearable Armband can Track User's Hand
06mApproaching the Real-World: Supporting Activity Recognition Training with Virtual IMU Data
06mBodyCompass: Monitoring Sleep Posture with Wireless Signals
22madjust.me! - AI-Based Table Auto Positiong