Event

PlaCo Bipedal Gait: ZMP-Constrained Trajectory Optimization and Weighted Inverse Kinematics

1 reports · 1 sources

GuYueJu analyzed the bipedal gait pipeline of the PlaCo humanoid robot: using jerk as the decision variable, it performs trajectory optimization on a linear inverted pendulum, constraining the ZMP within the support polygon; the swing foot trajectory is connected using cubic splines; then weighted QP inverse kinematics is used to map the center of mass and both foot targets into 26-dimensional joint velocities, thereby achieving continuous turning walking. This analysis breaks PlaCo's gait into two layers: trajectory optimization and inverse kinematics. The upper layer is responsible for generating center-of-mass and foothold trajectories that satisfy the ZMP constraint, and the lower layer is responsible for converting trajectory targets into joint velocity commands. The above description comes from GuYueJu's analysis and is not an effect validation published by the PlaCo team.