The concept of Smart Homes endeavours to allow senior citizens
to maintain their dignity without compromising their daily lives
and to help them to preserve an independent lifestyle. It enables
them to continue to live in their own homes for longer without
requiring live-in help or expensive institutional care.
To this end, we envisage that an intelligent human behaviour
recognition system installed in the home plays a constructive
and supportive role by monitoring a person"s movements in a
non-invasive manner, and determining when assistance is required.
In order to accomplish this objective several aspects of in-home
monitoring must be solved. The first challenge is to identify and
label objects in the scene. This is a prerequisite for recognising
high level behaviours since behaviours generally involve the
manipulation of objects. Unfortunately, traditional object recognition
techniques based on the visual appearance of an object are not reliable,
especially in cluttered environments. Hence, we are developing new
techniques for robust and incremental object recognition.
The second challenge is to build up a profile of the occupant"s
typical daily behaviours in the home. This can be used to detect
anomalous behaviours that may be indicative of emergency situations
requiring external intervention. Refinements on this technology would
facilitate the creation of "reminder" systems that are capable of
providing appropriate assistance without requiring outside aid.
To achieve the above aims we need to design software that accepts
input from multiple interconnected sensors, including cameras,
microphones, RFID tags, motion detectors, infrared sensors,
pressure pads, etc.
 |
 |
 |