Requirements for getting a robot to grow up

Much of current robot research is about learning tasks in which the task to be achieved is pre-specified, a suitable technology for the task is chosen and the learning process is then experimentally investigated. In this paper we discuss a different kind of problem: how to get a robot to 'grow up' through experience of its world. We discuss what this means and what it seems to require in a robot architecture. The main contribution of this paper is to suggest a particular research agenda for the community.