In the design and development of a legged robot, many factors need to be co
nsidered. As a consequence, creating a legged robot that can efficiently an
d autonomously negotiate a wide range of terrains is a challenging task. Ma
ny researchers working in the area of legged robotics have traditionally lo
oked towards the natural world for inspiration and solutions, reasoning tha
t these evolutionary solutions are appropriate and effective because they h
ave passed the hard tests for survival over time and generations. This pape
r reports the use of genetically inspired learning strategies, commonly ref
erred to as genetic algorithms, as an evolutionary design tool for improvin
g the design and performance of an algorithm for controlling the leg steppi
ng sequences of a walking robot. The paper presents a specific case of find
ing optimal walking gaits for an eight-legged robot called Robug IV and sim
ulated results are provided.