People involved from BSS
2015 – 2019
Many high-potential applications of daily-life monitoring (e.g., in telemedicine, virtual reality, and serious gaming) require 3D full body motion tracking. Unconstrained body movement capture anytime/anywhere requires the sensing system to be self-contained, unobtrusive, easy to wear, cheap and low-power. Existing systems for unconstrained movement capture require a sensor system to be strapped to each body segment to be tracked (for example for full body motion capture 17 sensors are needed), resulting in a relatively complex and expensive system.
The MiniSens project aims at accurately estimating unconstrained 3D full body movements using a minimal on-body sensor set, e.g., 4 or 5 sensors for full body capture. Project partners for this project are helping out with the sensor technology and/or applying the developed technology for their products. Applications one can think of are virtual reality (moving around (realistically) in a virtual environment, while wearing minimal sensing), healthcare (provide a product patients can use in their home environment to work on their training exercises) or maybe even consumer applications since wrist-worn activity monitors are becoming more mainstream this might be the next step.
The reduction in the number of sensors is achieved by using a database of movements measured with the IMU-based motion capture system of Xsens. Using this database we can build a model that maps the lower-dimensional input (reduced sensor set) to the higher-dimensional output (full-body sensor set).
For this project knowledge of different fields is used such as: signal processing, machine learning and information theory. The sensor technology is provided by Xsens, which consists of miniature MEMS inertial sensors.