Spatial Context from 3D Perception

Name: Boris Takac
Advisors: Prof. Dr. Matthias Rauterberg, Dr. Wei Chen, Prof. Dr. Andreu Catala1
Funding: Erasmus Mundus Joint Doctorate (EMJD) in Interactive and Cognitive Environments (ICE)), which is funded via the Education, Audiovisual and Culture Executive Agency (EACEA) of the European Commission under EMJD ICE FPA nº 2010-0012.

Duration: January 2011 – January 2014

Parkinson’s disease (PD) is a chronic progressive neurodegenerative disease involving gradual loss of motor and non-motor function. he motor symptoms of Parkinson’s disease include rigidity, tremor, bradykinesia (slow movement), postural alterations, tendency to fall, reduced gait speed, reduced step length, and episodes called freezing of gait (FOG), which consists of a blockade of the motor activity, resulting in a sudden inability to start or continue walking . The occurrence of freezing of gait phenomena is very dependent on situational and locational contextual aspects of the affected person, such as performing turns around obstacles, going through narrow passages or  reaching desired destinations. 

The technological advancement of the last decade, along with the necessity for more objective assessment methods resulted with the emergence of wearable PD monitoring systems. These systems usually employ inertial sensors to directly infer physiological context of the user, but they are not very well equipped for inference of other types of contextual information.

In our project, we focus on the inference of spatial context of Parkinson's disease patient. We consider that the spatial context of PD patients consists of their location in the room, their current movement action and their current distance to different types of obstacles.

Recent availability of novel camera systems providing both color and depth images (e.g. Microsoft Kinect), enables the use of 3D perception for tracking of people and direct extraction of geometrical properties of the observed environment. Our system is envisioned to be  a modular component that will augment wearable PD monitoring system with additional contextual information.

 

 

Primary components:

  • Localization module giving pose of the user on 2D floor map
  • Scene Model module employing geometrical analysis of 3D model of the environment and providing active 2D map for localization
  • Spatial Context Interpreter module aggregating user location and extracted properties of environment
  • High level data fusion with wearable system

Output

Takac, B., Cabestany, J., Catala A., Chen W., & Rauterberg, M. (2012). "A System for Inference of Spatial Context of Parkinson’s Disease Patients". Accepted by pHealt 2012.

Collaboration

Erasmus Mundus Joint Doctorate (EMJD) in Interactive and Cognitive Environments (ICE).

1CETpD - Technical Research Center for Dependency Care and Autonomous Living, Politechnic University of Catalonia, Barcelona, Spain