The sensorless case

In the case of having no observations, the path through ${\cal I}_{prob}$ becomes predictable. Suppose that a feasible planning problem is formulated. For example, there are complicated constraints on the probability distributions over $X$ that are permitted during the execution of the plan. Since ${\vec{X}}= {\cal I}_{prob}$ is a continuous space, it is tempting to apply motion planning techniques from Chapter 5 to find a successful path. The adaptation of such techniques may be possible, but they must be formulated to use actions and state transition functions, which was not done in Chapter 5. Such adaptations of these methods, however, will be covered in Chapter 14. They could be applied to this problem to search the I-space and produce a sequence of actions that traverses it while satisfying hard constraints on the probabilities.