ADAMAAS

Overview

Project duration

01.05.2015 – 31.03.2018

Funded by

Project website

Short description

ADAMAAS combines techniques from human memory research, eye tracking and physiological measurements (such as breathing rate or heart rate), object and action recognition (Computer Vision), Augmented Reality (AR), and advanced cognitive assessment and intervention techniques. The system is able to automatically detect the current state of an assembly task and provide individualized feedback at an appropriate time point.

Partners

Description

To help people suffering on dementia and the whole family

ADAMAAS combines techniques from human memory research, eye tracking and physiological measurements (such as breathing rate or heart rate), object and action recognition (Computer Vision), Augmented Reality (AR), and advanced cognitive assessment and intervention techniques. The ADAMAAS Glasses are able to identify actual action processes, to anticipate mistakes and react when errors are made, as well as to display situation- and context-dependent assistance in textual, 2D and 3D in-situ format superimposed on a transparent virtual plane in users' field of view. ADAMAAS aims to provide support for people to be able to live a self-sufficient life in an age-appropriate way according to their mental and physical capabilities. Thus, the system is able to suggest new action options and to induce selective learning processes. ADAMAAS is therefore not a stationary, but a mobile assistance system. The ADAMAAS glasses are able to measure participants gaze on the virtual plane in user's field of view as well as in the real environment. We designed a video-based, real time deep neural network architecture consisting of three modules. The first module can detect local hand- and object interaction using a semantic labeling approach. This module can track the orientation and position of objects in 3d space using model based registration. The second module estimates the global state of an assembly task. Combined  with the third module, a probabilistic state machine, the system is able to automatically detect the current state of an assembly task and provide individualized feedback at an appropriate time point.

In 2018, the initiative "Land of Ideas", founded by the German government and the Federation of German Industries, praised ADAMAAS (“Adaptive and Mobile Action Assistance in Daily Living Activities”) as an innovative and forward-looking project that benefits the country and its inhabitants, and honored it as one of 100 "Landmarks in the Land of Ideas" (out of approx. 1.500 candidates).

Impressions & Documents