Hands-On ROS for Robotics Programming

Observations

Given an environment, an observation consists of a set of values that define a given state of the environment, the angle θ. It is like taking a snapshot of a scene. The environment's step function, env.step, returns the following:

• The current state; that is to say, it sets the current value of the state variable, θ. Operationally, it is an object type variable called observation.

• The reward that the agent has obtained from the last action, force F, as was previously mentioned when describing the action space. The reward is like points in a game—a quantitative value that accumulates all the rewards (points) obtained from the actions performed in the current episode from the beginning, that is, the current score in the game ...