Seminal works in biped walking such as [1] had assumed favorable initial conditions on a limit cycle to start with. A more practical approach would be to start the biped from a static resting position. This paper proposes a simplified and straightforward approach for taking a biped robot from an initial resting position to a stable walking limit cycle. The biped model selected for the study is the 'kneeless biped with torso'. The problem is tackled in two stages - gait initiation from rest followed by convergence to stable walking. Walking is divided into various sub-phases depending on the state of the biped and simple state feedback control laws are proposed for each phase. The individual control laws are mathematically accrued into a single control law valid throughout the walking phase. Simulation results validating the approach are presented.