Most researchers study biologically inspired bipedal robot by implementing sophisticated control algorithms into a mechanical structure with actuators and stiff or compliant joints, or simplifying robot based on passive dynamics of the body. However, few study focused on the biomechanical view of the musculoskeletal system of human body, namely, the morphological computation of human body. In this thesis, I propose a new approach of developing bipedal robot inspired from human musculoskeletal (MSK) biomechanics. A three-dimensional (3D) whole-body musculoskeletal model was used as the biological counterpart to inspire the design of the robot. Image processing was used to examine the anatomy, structure and geometry of human MSK system. Reverse engineering was applied to rebuild the 3D musculoskeletal model. With the model, the details of the skeletal and the muscular system were quantified and modified. The anthropometric data, the mechanical properties and muscle arrangements were analysed in the model. Key kinematic parameters were obtained from 3D motion capture system and inverse dynamics technique. Based on analysis of the body structure and the mechanics of human MSK, Computer-aid design (CAD) and 3D printing technique were adopted to design and manufacture the skeletal body of the robot, assuring the mechanics in accordance with real human body. Human muscles are not isolated from each other, in fact, they are linked by fascia or even connected to the same tendon, e.g., Soleus muscle joins with Gastrocnemius muscles to form Achilles tendon which inserts onto the posterior surface of the heel bone. Therefore, I designed a whole-body muscular system in which the artificial muscle units with similar function are woven into textiles, especially for Soleus and Gastrocnemius because they are the most important muscles during bipedal locomotion. Computer simulation of passive walking of the robot was conducted in Adams. The kinematic and kinetic data were measured to compare with humans. The effect of the ankle orientation on normal walking was studied using the design of experiment. The best configuration in the ankle is 16Â° talocrural angle and 23Â° subtalar angle, where the robot can travel on the ramp up to 4166.2 mm, 17.4 % longer than the distance that the robot can travel with 0Â° talocrural angle and 0Â° subtalar angle (parallel to the ground when standing). It can be intuitive thinking that the oblique axis of rotation in the ankle may facilitate normal walking for biped robots. Moreover, physical tests getting down a 2.44 m ramp were carried out. The robot with 0Â° talocrural and subtalar angle can passively walk the full length of the ramp in about 95% of launches, whereas inappropriate initial conditions, e.g., slope angle, released roll angle and released velocity, seem to be the primary cause of those launches in which the robot stopped or fell down before reaching the end of the ramp. Successful walking of the robot validated the feasibility of the new robotic developing philosophy. The presented framework would provide a biomechanics foundation and technical support for the innovative design and manufacture of bio-inspired bipedal walking robot. Furthermore, this paradigm for developing bipedal robot can be also considered as a new approach to develop exoskeletons and prostheses based on human musculoskeletal system.