Imager based sensor networks for understanding and creating behaviors

In this vision paper we propose the creation of sensor networks based on custom designed multifunction imagers that can directly pick out events of interest from a scene. We argue that with a suitable learning framework such sensors have great potential for realizing sensor networks that can interpret and subsequently orchestrate behaviors in physical space. This paper provides an overview of the research challenges followed by a brief description of our current research efforts towards creating a functional experimental system for investigating these intelligent networks