design a reinforcement learning system to control traffic lights.

Imagine you have a crossroad to control.

Two traffic light sets are installed perpendicular to each other.

You can use the penalty term as:

Accident: any car accident penalises the system drastically

Delay: the average time cars need to completely pass the crossroad is a direct penalty term

Develop a reinforcement learning algorithm.

$1)$ Define penalty terms

$2)$ Identify sensors that would be needed

$3)$ Design an algorithm for the reinforcement system