This paper presents a method for generating intelligent upright biped stepping motions for real-time dynamic environments. Our approach extends the inverted pendulum (IP) model by means of an impulse-based technique to achieve rigid-leg constraints during foot support transitions. The impulse-based method in cooperation with the IP method provides a computationally fast, straightforward, and robust solution for achieving stiff-knee joints that are desired during casual stepping motions, such as standing and walking. Furthermore, we demonstrate how the impulse-based inverted pendulum (IIP) model can be extended to embody rotational information to synthesize more dynamic actions, such as when the feet leave the ground or when slipping (i. e., foot friction).