Case Studies

The goal of the following case studies is an introduction into those theories and algorithms which enable the learning of patterns of movements of people living in a flat. The learning should be located in an intelligent agent constructed within the general assumptions of the $GB^{2}CLT$. To support learning one has to provide a minimal simulator for the generation of appropriate data. For the usage of the intelligent system in connection with a real world scenario the simulated data have to be calibrated with real world measurements. Within this case study the goal is limited to the learning of simulated data. For more details see the text below.

Figure 8.1: AAL SmartEnvironment minimal components

The context of this project is given by the AAL smartEnvironment research program of the AAL Lab of the Masterprogram Barrierfree Systems of the University of Applied Sciences. (See also the URL http://www.uffmm.org/aallab). An overview of the main components can be seen in the figure 8.1. Within this overall scenario there is one possible subtopic Emergency Care. This includes all kinds of technology which enable the detecction of critical situations for persons of a target group. The arget group here shall be elderly people which still live at home although there may be some deficiencies in the bodily or cognitive capabilities.

A special focus will be on those environments, where a Sens Floor^8.1 is part of the flats under investigation. The number and types of human persons living in a flat with a Sens Floor can vary remarkably.

In the following paragraphs we will investigate some formal methods how to model learning systems having as input data those of a flat sensor field as well as controlling a moving object in the space above this field. For to do this we will define a set of test cases which will be used for different formal approaches. The main goal ist to test and further elaborate the concept of an artificial consciousness. Some of the topics to be answered are the transition from points in space to features and then from features to categories including static structures as well as dynamic ones.