Overview / Usage

In this research, an animation was designed to determine an appropriate physical setup for depth sensors in an institutional hallway, for human foot surface capture and tracking. The sensor configuration was tested by creating the setup in a laboratory with Intel RealSense D415 depth cameras, synchronizing multiple depth sensors (spatially and temporal), collecting depth camera data, and generating a point cloud of a person walking in the capture area. Simulations showed that 96% of a walking human’s foot surface can be reconstructed from depth sensors data, using sets of four diagonal corner locations for each instance.

The revamp of the system will require tighter integration of the systems and use of Intel solutions to accelerate AI and ML.

Methodology / Approach

This project is in close collaboration with researchers at the university of Ottawa.