An alternative approach to synthesizing bipedal walking

Area of Science:

• Robotics
• Biomechanics
• Control Engineering

Background:

• Humanoid bipedal robots require sophisticated control systems for stable locomotion.
• Existing models often involve numerous parameters, complicating gait generation.

Purpose of the Study:

• To develop a simplified yet effective control scheme for generating cyclic gait in a seven-link humanoid biped.
• To identify essential gait descriptors for stable bipedal walking.

Main Methods:

• Mechanical analysis to determine critical gait descriptors: step length, step time, and center of mass (CoM) velocity at push-off.
• Model predictive control (MPC) using quadratic dynamic matrix control to regulate gait descriptors as end-point conditions.
• Implementation of continuous controls for trunk uprightness and weight bearing.

Main Results:

• Identified step length, step time, and CoM push-off velocity as key controllable gait parameters.
• Demonstrated that two descriptors can be independently controlled due to ballistic CoM motion during swing phase.
• The MPC successfully generated symmetric and stable gaits by specifying step length and CoM push-off velocity.

Conclusions:

• The proposed control scheme offers a general-purpose solution for bipedal gait generation with fewer, directly relevant parameters.
• The model's parameters (step length, trunk orientation, step time, walking velocity, weight bearing) are intuitive determinants of gait.
• Potential applications in advanced robotics and rehabilitation engineering for humanoid locomotion.