Evolutionary Robotics is an expanding area in the world of robotics that incorporates ideas from fields such as Biology, Engineering, and Computer Science. The main idea behind Evolutionary Robotics is to use computerized models of biological evolutionary phenomena to evolve robotic behaviors. In recent years, new breakthroughs have been made to further its development, prove its effectiveness, and provide solutions to interesting problems within the robotics world. The approach of evolving controllers for autonomous robots has established benefits over a more traditional hand-­‐coded approach. Like most fields of research, Evolutionary Robotics contains its own set of problems. One such problem involves the use of simulators to speed up the evolutionary processes. When transferring the robotic controller from the simulation to the physical robot its performance tends to decrease on a given task; this issue is referred to as the reality gap problem. In this research, a new approach to bridging the reality gap is presented and explored. The idea is to evolve a robotic controller that generates desires based on its current state and uses reinforcement learning to select actions that achieve these desires. By doing so, the goal is to have a robotic controller adapt to differences, uncertainties, and perturbations within the real world once transferred from simulation.