Some lessons learned on constructing an automated testbench for evolvable hardware experiments

A testbench is absolutely essential to conducting evolvable hardware research, particularly if intrinsic investigations are conducted. Although testbench design is straightforward, in practice its construction has numerous pitfalls. This work describes the automated testbench used in the Adaptive Systems Laboratory at Portland State University. We identify the problems we encountered in making the testbench operational and how those problems were resolved. Finally, we offer some guidelines others find useful when constructing their own evolvable hardware testbenches.