Long range UHF RFID technology can be used to detect object presence, measure their location relative to antenna position, and read information such as object type, profile and preference that could be stored right on the tags. Today, novel RFID tags are also capable of sensing environmental information. These information can be used to understand an objects context, and the context information coming from multiple objects can be used to determine the “situation” governing the environment (e.g. room) where objects are situated.This kind of situation awareness is a key attribute to Ambient Intelligence.
As an example scenario, consider a classroom where we want to detect situations such as “setup”, “delivering the lecture”, “coffee break”, “question and answer session”, “exam”, “close up” etc.
One key challenge to situation recognition is the computational complexity when dealing with multidimensional context data. In particular, because in Ambient Intelligence the recognition is supposed to get done by resource-limited computers that are arranged in a de-central architecture and co-operate with a partial topology knowledge.
The idea to overcome this problem is to use a uni-dimensional context information coming from a very large amount of objects. Imagine having the presence information from 100-1000 objects that are among the “frequently visiting objects” of a classroom: ranging from your pen, books, computers, jacket, “Tschador”, bags, shoos to those objects owned by the university such as tables, projectors, seats etc.
UHF RFID tags can be massively deployed to label all kind of everyday objects that could be detected using long-range RFID sensors. Such sensors can scan more than 500 tags per scan. One or two scans per second is a realistic operation scenario. A globally unique ID such as the EPC represents an object. Although RFID tags are capable of storing additional information such as, in this research we are interested to explore whether or not just the presence information can allow for a robust situation recognition. If the research successful, we can construct such environments using only one type of sensors: UHF RFID. In contrast to other kind of sensing technology, RFID is robust, widely accepted, and available. It is a off-the-shelf product. Worldwide, a large industry is working on improving it around the clock. Training and certification programs for system engineers to deploy and maintain RFID is widely available. Thus, using RFID technology to construct Ambient Intelligence environments will reduce acquisition, operation, and maintenance costs.
The research question is how robust one can recognize situations in an intelligent classroom using a single type of object presence sensor.
Note: both hardware and software with example source code are available in the lab.