a similar approach and uses partial planning and delayed decisions. 
All those approaches have in common that, in case of a failure, they have to either change the beliefs, locate the fault, or gather additional information about the environment to choose a (new) plan. In contrast, our approach does not have to locate the fault but rather uses statistics to locate the plan's reliable actions. In particular, the calculation with SFL is very inexpensive and can be adapted for different approaches. Even extending former mentioned approaches should be possible as long as we can integrate a cost function into the planning algorithm. To the best of our knowledge, we do not know of a similar approach for replanning.