Evolutionary robotics - as other adaptive methods, such as reinforcement learning and learning classifier systems - can take considerable time and resources which require a careful evaluation of the hardware tools and methodologies employed. We outline a set of hardware solutions and working methodologies that can be used for successfully implementing and extending the evolutionary approach to complex environments, robots, and real-world applications. The issues discussed include the integration of simulation and real robots, design issues of evolvable robots, hardware requirements for incremental evolution, and hardware and software tools for monitoring and analysis.