1: \begin{abstract}

2: Modern industrial applications require robots to be able to operate in unpredictable environments,

3: %It is also desirable that new robot policies or 

4: and programs to be created with a minimal effort, as there may be frequent changes to the task. In this paper, we show that genetic programming can be effectively used to learn the structure of a behavior tree (BT) to solve a robotic task in an unpredictable environment. Moreover, we propose to use a simple simulator for the learning and demonstrate that the learned BTs can solve the same task in a realistic simulator, reaching convergence without the need for task specific heuristics. %This is achieved by synthesizing BTs which are naturally reactive to failures and including the success probability in the fitness function, driving towards solutions that avoid actions that are more likely to fail. Finally, we show that

5: The learned solution is tolerant to faults%on the manipulation, localization and navigation actions

6: , making our method appealing for real robotic applications. 

7: \end{abstract}

8: