A 'Virtual Presence' demonstrator has been created for the Metaverse1 project, by Almende, Salland electronics and the TU/Eindhoven. The use case 'virtual presence' is concerned with creating a seamless integration between the real and virtual world by using real and virtual sensors and actuators. The use case focuses on independent assisted living through non intrusive monitoring of activities of daily life, and by giving means to easily communicate with the outside (real and virtual) world.

For the demonstrator we created a virtual representation of the 'Context Lab' at the TU Eindhoven in the open source Second Life software OpenSim.





This virtual environment can be visited on our demo server after creating an account here.

With the use of a device service framework, created for this use case, a connection was made between the real devices situated in the physical room and the virtual devices in the virtual replica. For the demonstrator the physical room was equipped with an Icat: and a wireless sensor network (WSN) of myriaNodes and smartphones. The Icat served as communication channel between both worlds. From the real world a video stream could be initiated which was displayed in the virtual world. The video stream coming from the Icat could be steered with commands from the virtual world as, 'icat look left' and 'icat look right', and via text-to-speech the physical Icat could say text entered in the virtual world.



The WSN with its sensors was used for detecting: presence, chair occupation, ambient light, human posture and a falling person. With this information simple activities in the real environment could be displayed in the virtual world. For instance if a person with a wireless node would enter the real room and switch on the lights, his avatar would be displayed and the lights in the virtual room would go on. Then if the person would sit on a chair with a pressure sensor his avatar would also sit on the corresponding chair, and if the seat would be occupied for a suspicious long time, the Icat would try to get a response from the person and otherwise contact his family via the phone or virtual world.

A more complex pattern that could be sensed was the posture of a person. The complexity of the posture patterns that could be recognized depended on the amount of acceleration sensors used on the body of the person. For a person carrying for instance only one smartphone only postures as sitting and standing (walking) can be recognized, whereas for a person with three MyriaNodes on the lower, middle and upper parts of his body, also different types of postures could be recognized with respect to different orientations of these body parts.

By combining fall detection with posture recognition the amount of false positives for fall detection can be reduced and the severity of a fall can be estimated by the posture. This was used in the demonstrator for sending out an alert phone call for help when a person would remain on the ground after he fell (see the video below).