A new walking pattern classification method is proposed for uneven floor walking of 5-link 7 DOF biped robot. This method extracts the patterns as per the stance foot's current floor position and swing foot's transitioning floor conditions during locomotion. When a global path composed of stairs, obstacles, etc. and certain walking parameters, such as the speed of walking and total walking time, are given to the system, the guidance controller unit determines footstep trajectory in terms of step patterns by using a genetic algorithm based optimization technique while ensuring biped's stability criterion. The demonstration of biped for different pattern classes is realized by a dynamic simulator.