In designing a new generation of legged robots, it is critical to understand the design principles employed by animals. One of the key steps to successful development of such bio-inspired robots is to systematically extract relevant biological principles, rather than direct copying features of an animal solution, which may be impossible to realize or irrelevant in engineering domain. The talk will introduce several examples that successfully implement bio-inspired design principles learned from animals. Our highlighting example is the development of  the MIT Cheetah, currently running at 13.5mph with an locomotion efficiency rivaling animals. Three research thrusts of the MIT Cheetah are optimum actuator design, biotensegrity structure design, and the momentum balancing control architecture for a fast and stable gallop. Each research component is guided by the biomechanics studies of runners such as dogs and cheetahs capable of fast traverse on rough and unstructured terrains. Through this project, we seek to derive design principles of quadrupedal locomotion that share characteristics with available mechanical and electrical capabilities in order to develop most efficient, robust robots, which will be part of our life in the future.