Ultrasonic Head-Pose Estimation

Amazon Lab126 Engineering, 2022–23

Liaison(s): Trausti Kristjansson, Patrick Hegarty, Krishna Kamath
Advisor(s): Philip Cha
Students(s): Alec Vercruysse (TL-F), Jeremy Bakken (TL-S), Declan O’Neill, Cole Nagata, Kevin Wan, Tristan Huang, Eli Rejto

Many traditional head-tracking methods require a clean visual environment or a physical IMU device, which can limit commercial head tracking applications. The Amazon Lab126 Clinic team is working to solve the head-tracking problem with ultrasonic sensing. We have developed a platform where ultrasound is emitted from a device, reflected off surfaces and received by on-board microphones. The signals are then input into an ML model that can predict the position and rotation angle of the head.