Walking controllers for humanoid robots with position-controlled joints
I developed walking controllers for humanoid robots with position-controlled joints in Robotics and Mechanisms Laboratory (RoMeLa) at Virginia Tech. The controller gets the target walking velocity as an input and outputs target joint angles to the servo motors. In detail, 1) reference pelvis and feet trajectories are calculated for a target velocity based on the linear inverted pendulum model (LIPM) to keep the zero moment point (ZMP) under the ankle; 2) these reference trajectories are modified based on the IMU sensory data; and 3) inverse kinematics is solved to calculate the target joint angles.