Inspired by animals ability to learn and adapt to changes in their environment during life, hybrid evolutionary algorithms which include local optimization between generations have been developed and successfully applied in a number of research areas. Despite the possible benefits this kind of algorithm could have in the field of evolutionary robotics, very little research has been done on this topic. This thesis explores the effects of learning used in cooperation with a genetic algorithm to evolve control system parameters for a fixedmorphology robot, where learning corresponds to the application of a local search algorithm on individuals during evolution. Two types of lifetime learning were implemented and tested, i.e. Baldwinian and Lamarckian learning. On the direct results from evolution, Lamarckian learning showed promising results, with a significant increase in final fitness compared with the results from evolution without learning. Machine learning is sometimes used to reduce the reality gap between performance in simulation and the real world. Based on the possibility that individuals evolved with Baldwinian learning can develop a potential to learn, this thesis also examines if learning could be advantageous when such a method is used. On this topic, the results obtained in this thesis showed promise in some sample sets, but were inconclusive in others. In order to conclude in this matter, a larger quantity of samples would be necessary.