Symbol q

The passage around this formula

…torque to fight the very contact it should be responding to. An impedance controller instead makes the joint behave like a programmable spring and damper around the commanded target: $\tau = K_p (q_d - q) + K_d (\dot{q}_d - \dot{q}) + \tau_{ff}$. Here $q_d$ and q are the desired and measured joint positions, $K_p$ and $K_d$ set how stiff and how damped that virtual spring-and-damper is, and $\tau_{ff}$ is a feedforward term — gravity compensation, or a reaction-force term handed down from a higher-level controller — added on top. The learned policy’s…

Read this part in the article →