We present a novel human-aware epistemic planning framework designed for collaborative human-robot interactions, specially tailored for situations where the agents’ shared execution experiences can be interrupted by the uncontrollable nature of humans. Our objective is to generate a robot policy that accounts for such uncontrollable behaviors, thus enabling the anticipation of potential progress achieved by the robot when the experience is not shared, e.g., when humans are briefly absent from the shared environment to complete a subtask. But this anticipation is considered from the perspective of humans who keep an estimated robot’s model. As a first step to address it, we propose a general planning framework and build a solver based on AND/OR search which integrates knowledge reasoning; this includes assessing situations by perspective taking. Our approach dynamically models and manages the expansion or contraction of potential worlds while tracking whether or not agents share the task execution experiences. This helps the planner (or the robot) to prepare itself with a set of worlds that humans would consider possible. The robot assesses the situation from the human perspective and removes the worlds that it has reason to think are impossible. However, there might still be an impossible world that is indistinguishable from the real world. In different situations, thanks to our planning framework, the robot’s policy built offline can determine an appropriate course of action, such as answering human queries, explicitly communicating some fact without being annoying, or taking appropriate action in the presence of the human to help them narrow down the possibilities further, facilitating collaboration. Our preliminary experiments show that the framework is effective for behavior planning in different situations. We discuss the practical issues in different problem settings.