These days, research on humanoid robots has made rapid progress for dexterous motions along with hardware development. Various humanoid robots have demonstrated stable walking algorithms. In most walking pattern generation methods for humanoid robots, it is assumed that the terrain is flat. However, in human environments, there exist inclined terrain, stairs, and uneven terrain as well as flat terrain. Therefore, walking pattern generation on various environments for humanoid robots is one of the key research issues.
In this thesis, an extended modifiable walking pattern generator (MWPG) is proposed for modifiable walking of humanoid robots on various environments. The humanoid robots in the previous researches related to walking pattern generation on inclined terrain, stairs, and uneven terrain were unable to independently modify the elements of a walking pattern, i.e. the single and double support times, the sagittal and lateral step lengths, the foot height, and the foot direction of the swing leg, without any extra footstep for adjusting the center of mass (CoM) motion. Moreover, only the inclination along the pitch direction was considered for walking. In real environments, however, there exist inclined terrains in roll as well as pitch directions. To solve these problems, the MWPG is extended, which allows the zero moment point (ZMP) variation in real-time by closed form functions. The MWPG can independently modify the elements of the walking pattern without any extra footstep for adjusting the CoM motion. However, it can be applied only on flat terrain. Thus, in this thesis, an extended MWPG is developed to independently modify the elements of the walking pattern on inclined terrain, stairs, and uneven terrain.
In the MWPG, to generate a walking pattern on flat terrain, a command state (CS) was defined as a navigational command set of the single and double support times, the sagittal and lateral step lengths, and the foot direction of the swing leg. I...