Used RxJS to create an observable in the ant model which triggers an action intent (for now, just "move" and "turn"). The environment subscribes to the intent determines the result of the ant's action. If the action is successful, e.g. the ant moves forward, a corresponding positive response is returned. If the action is unsuccessful, e.g. reached the edge of the grid, a negative response is returned. These responses will be used to train the neural network and allow the ant to learn from its environment.