A Inverted Pendulum Details

A.1 Inverted Pendulum Model and Controller The inverted pendulum system is modeled using state s =[,!] where is the angle of the pendulum from the vertical, and! is its angular velocity. We clip! such that the magnitude of the angular velocity does not exceed 8 rad/s, and we clip the control inputs so that the maximum torque magnitude does not exceed 2 N m. We derive a simple rule-based policy to balance the pendulum according to the following two equations! The second equation performs proportional control using this quantity and the current angular velocity. This controller is able to keep the pendulum upright under perfect perception.