This work presents a biologically inspired method of gait generation. Bipedal gait pattern (for
hip and knee joints) was taken into account giving the reference trajectories in a learning
task. The four coupled oscillators were taught to generate the outputs similar to those in a
human gait. After applying the correction functions the obtained generation method was
validated using ZMP criterion. The formula suitable for real-time motion generation taking
into account the positioning errors was also formulated. The small real robot prototype was …