One of the long-term goals in evolutionary robotics is to be able to automatically syn-thesize controllers for real autonomous robots based only on a task specification. Whilea number of studies have shown the applicability of evolutionary robotics techniquesfor the synthesis of behavioral control, researchers have consistently been faced witha number of issues preventing the widespread adoption of evolutionary robotics forengineering purposes. In this article, we review and discuss the open issues in evo-lutionary robotics. Firstly, we analyze the benefits and challenges of simulation-basedevolution and subsequent deployment of controllers versus evolution on real robotichardware. Secondly, we discuss evolutionary computation-specific issues that haveplagued evolutionary robotics: (i) the bootstrap problem, (ii) deception, and (iii) therole of the genomic encoding and of the genotype-phenotype mapping in the evolution of controllers for complex tasks. Finally, we address the absence of standard re-search practices in the field. In addition to the review and discussion of the issues, wefurther discuss a number of promising avenues of research. Our underlying motiva-tion is the reduction of the current gap between evolutionary robotics and mainstreamrobotics, and the establishment of evolutionary robotics as a canonical approach for the engineering of autonomous robots.