Biorobotic Locomotion: Biped Humanoid Walking using Optimal Control
MetadataShow full item record
This paper explores the use of optimal control for quasi-static bipedal walking trajectory synthesis. Optimal control differs from the traditional method of inverse kinematics used for generating trajectories. On setting up an optimal control problem and solving it, a set of joint angle trajectories for executing a portion of the gait can be synthesized using the forward kinematic description, as opposed to the method of inverse kinematics. Although optimal control solution requires the same computational processing, the advantage is the simplification of problem set up procedure. This research aims at developing a quasi-static walking gait for a humanoid by defining a set of optimal control problems with constraints and costs defined as functions of joint angles. OPTRAGEN, a Matlab toolbox, is used to convert the optimal control problem to a non-linear problem which is then solved by a numerical optimization program. The goal is to generate a planar walking gait by setting up an optimal control problem that minimizes the change in joint angles while keeping the body stable. The generated planar gait was tested on the physical robot. On completion of this, the same methods were also used to generate trajectories for a 3-Dimensional quasi-static walking gait.