Kinect Shoulder Detection
NOTE: This project is under an additional NDA under TECNALIA. Release of all information is not available.
The Kinect Shoulder Detection project aims at developing software to be used in detecting and storing shoulder and torso position with accuracy and precision. The position data will be stored in an external database for patient records and analysis.
Problem Statement
Currently the ArmAssist system being used for therapy does not track torso or shoulder movement while a patient is using the rehab station. If the patient were to not be monitored while using the ArmAssist station, they could potentially cheat and hinder their therapy.
Summary of System Design
This document lays out the minimum requirements expected of our team from the customer.
Shoulder Detection Program (SDP)
- The following features are a must to solve the problem statement.
- Calibration to gather base data to be used when calculating an acceptable range of movement for the user.
- Collection of data during run-time relating to a specified shoulder as well as the torso to improve accuracy and precision in measurements.
- Storage of x, y, and z positional coordinates for all multiple tracking points during run-time that are recorded for each frame captured.
Kinect Fusion
This document lays out research done into the Kinect Fusion abilities of the Kinect. This function of the Kinect would be used to create a 3D model of what the Kinect sees and create a grid off of that so we can get information about the patient AND the ArmAssist station.
Kinect Data Point Testing
This document lays out research done into the Kinect's Data Points and what the returned values from the Kinect really mean. Most of the information is provided by Microsoft but for completeness and to give more than just a general overview, we tested every coordinate plane and observed our results.
Group Members
This project was complete on 12/13/13 by Alex Cochrane and Brett Papineau