With increasing requirements of exploring challenging areas like rugged and slippery environments, interest in legged robots also has been growing. Especially for operating the legged robot in various conditions, it is essential to estimate its state precisely even under slippery or vision denied environments. Therefore, this paper proposes a state estimator for a legged robot using additional leg kinematics information. For this, we introduce a foot velocity factor that calculates the relative foot poses and added it to the factor graph. Our algorithm is verified on several datasets, including slippery or feature-less environments.