The document is a master's thesis from February 10, 2009 on a biologically inspired learning method for biped rhythmic walking using static freezing and freeing. It proposes a learning method where degrees of freedom in a robot are temporarily frozen during early motion learning and then freed as learning progresses. This process is aimed at reducing difficulties in learning and converging on a solution. The method is examined through a numerical computer simulation of a biped walking robot learning to walk rhythmically. Figures 1 and 2 show snapshots of the robot walking with frozen and freed degrees of freedom, respectively. The research achievement section lists a conference paper by the author and others on learning rhythmic walking in a biped robot based on demonstrated motions.