trained_policy.txt

import numpy as np

def solve_cartpole(cart_position: float, cart_velocity: float, pole_angle: float, pole_angular_velocity: float) -> int:

    # Define the constants for the PD controller
    kp_pole = 0.5
    kd_pole = 0.1
    
    kp_cart = 0.2
    kd_cart = 0.05
    
    # Compute error signals and their derivatives
    pole_error = pole_angle
    pole_error_derivative = pole_angular_velocity
    
    cart_error = cart_position
    cart_error_derivative = cart_velocity
    
    # Compute the control signals for the pole and cart using PD controllers
    pole_control_signal = kp_pole * pole_error + kd_pole * pole_error_derivative
    cart_control_signal = kp_cart * cart_error + kd_cart * cart_error_derivative
    
    # Combine the control signals to get the final action
    if pole_control_signal < -cart_control_signal:
        action = 0
    elif pole_control_signal > cart_control_signal:
        action = 1
    else:
        action = 1
        
    return action