Fall is a major risk for elderly people. This paper is an outline of the research work that we are conducting in our group concerning the development of new technologies for fall detection and prevention at home. Our aim is to propose new affordable devices at home which: (1) automatically detect falls, and then alert whom is concern when a fall has been detected, (2) proceed with some measure in order to define an indicator associated with the risk of fall. Such devices could reassure persons affected by mobility problems or being recently injured in a fall, thus permitting them to stay at home longer. We are currently examining how low-cost RGB-D cameras could be used to track continuously a person at home. We show that we can easily extract, from depth images, the body center of mass of a person and some other simple parameters from which we can detect and prevent falls. Preliminary results are presented based on two real experimentations with young people, within an experimental smart home. 208 sequences were recorded for the first experimentation concerning fall detection and 106 strides were analyzed for gait parameters measurement.