Recently there has been a lot of focus on human robot co-habitation issues that are usually orthogonal to aspects of human-robot teaming; e.g. on producing socially acceptable behaviors of robots and de-conflicting plans of robots and humans in shared environments. However, an interesting offshoot of these settings that has largely been overlooked is the problem of planning for serendipity - i.e. planning for stigmergic collaboration without explicit commitments on agents in co-habitation. In this paper we formalize this notion of planning for serendipity for the first time, and provide an Integer Programming based solution for this planning paradigm. Further, we illustrate the different modes of this planning technique on a typical Urban Search and Rescue scenario and show a real-life implementation of the ideas on the Nao Robot.