Abstract:

In this paper, we present the outcomes and conclusions obtained by involving seniors from three countries (Denmark, Poland and Romania) in an innovative project funded under the European Ambient Assisted Living (ALL) program. CAMI stands for “Companion with Autonomously Mobile Interface” in “Artificially intelligent ecosystem for self-management and sustainable quality of life in AAL”.

The CAMI solution will enable flexible, scalable and individualised services that will support elderly to self-manage their daily life and prolong their involvement in the society (sharing knowledge, continue working, etc). This will also allow their informal caregivers (family and friends) to continue working and participating in society while caring for their loved ones. The solution is designed as an innovative architecture that allows for individualized, intelligent self-management which can be tailored to an individual’s preferences and needs. A combination of advanced reasoning algorithms with a voice-based interface is easing the self-management and decision making process. The decision support module is an expert system that acts as the central integration point for CAMI. It is a dockerized container, deployed in the cloud.

A user-centered approach has ranked health monitoring, physical exercises and voice based interaction among the top favoured CAMI functionalities. Respondents from three countries (Poland, Romania and Denmark) participated in a multinational survey and a conjoint analysis study. E-health solutions were perceived as very useful and 59% of the respondents considered the graphic display of various health measurements (e.g. blood pressure, heart rate, oxygen levels) as an interesting feature. The ability to share health measurements with various doctors was perceived as useful by 60% of the respondents. Computer supervised physical exercises ranked third in the conjoint analysis. All the voice control functionalities received more than 50% votes which classifies them as promising features.

The physical exercise module is based on two avatars: the training avatar and the avatar of the user. The training avatar is performing different physical exercises which have to be reproduced by the user. The user’s movements are compared with the movements of the avatar based on algorithms for comparing two nonlinear series. At the end of the exercise, the user receives a score that reflects the correctness of the performed exercises. The results are saved in a database, such that they can be analyzed by a caregiver or by a medical specialist.

The voice interface is composed of five main parts: Automatic Speech Recognition, Natural Language Understanding, Dialog Management module, Natural Language Generation, and Text to Speech synthesis. Since, the user should always be able to interact with the CAMI platform, regardless of the status of the internet connection, the vocal interface should work also offline ensuring basic system functionalities. Therefore, the vocal interface will have two working modes:
(1) an online mode that depends on internet connectivity and
(2) an offline mode (a limited version) that doesn’t depend on internet connection.

Keywords:

User-centred design, fall detection, physical exercises, vocal interface.