I. Introduction
Transcending industrial contexts, robots increasingly enter public and private environments. Robots are typically deployed to support humans in diverse tasks or to relieve them of tasks altogether. In fiction and marketing, robots often act fully autonomous. Consequentially, unrealistic expectations about currently feasible capabilities of robots are present in the general populace. However, tasks exist, which robots cannot solve easily. An example is the task to open doors. Purely technical solutions of using manipulators or building automation can be coupled at large cost. Even when manipulators are available, many tasks are non-trivial and specialized end-effectors are necessary. Specialization however can be problematic when confronted with an unexpected situation, such as an unknown type of door opening mechanism. Robots often face difficulties when operating in dynamic environments, since they are lacking in flexibility to adapt to new conditions. As humans are often around in environments made for them, and as they can act competently within them, an obvious approach is to ask them for help. If the capabilities required for achieving help from humans are less complex or costly than for solving the actual problem, integrating human help into the solution becomes feasible.