• UFRJ Nautilus

Artificial Intelligence - from video games to AUVs

Updated: May 7

In most video games there is an opponent, in multiplayer games you play against another human, but in single players the opponent is the “computer”. This opponent is commanded, usually by an artificial intelligence (AI), object of a great study by the game programmers and, therefore, of great improvement in the last years. In addition to electronic games, another technology that uses AI is that of AUV (Autonomous Underwater Vehicles).


Artificial intelligence allows machines to "think" like humans, independently and based on digital data. Any device without people control needs some form of AI, so it is widely used for various media, including games.


In older games, AI was simpler. In Pac-Man, released in 1980, for example, there are 4 types of ghosts with 4 different colors, red, pink, blue and orange. The red works in the least complex way among the 4, since it only follows the Pac-Man, the pink, on the other hand, tries to close the main character, the orange gives up when the user moves away, while the blue is based on the positions of Pac-Man and red to define its direction.


Jogo "Pac-Man". UFRJ Nautilus
Pac-Man game.

The most common way of representing the behavior of an Artificial Intelligence in non-player characters (NPC) is a finite State Machine. The State Machine is formed by a set of states, consisting of actions that will be taken, and transitions, which are the rules for changing them. The image below shows a State Machine from an FPS (First Person Shooter) game. But how does this relate to the behavior of an AUV?

Máquina de Estados de um jogo FPS (First Person Shooter). UFRJ Nautilus
State machine of a FPS game (First Person Shooter).

Considering that all AUV is autonomous, it needs Artificial Intelligence to carry out its decision-making without external interventions, and ours is not different. Our robot uses AI in the areas of neural network, in location and mapping algorithms and in control. In addition, due to the completely different tests within the competition, a State Machine is used, which is a form of AI. To implement the State Machine in the robot, we use Smach, a ROS library that allows the implementation of this technology more easily. Our State Machine works in layers: a larger one with a state for each competition event and within each state there is another machine to perform each event specifically.


Therefore, it can be seen that, although AUVs and video games seem to have completely different themes, they have more in common than expected. Thus, we verify the importance of interdisciplinarity for the construction of an AUV and how all knowledge is useful and important and how Artificial Intelligence is relevant to present and future technological progress.


Written by Thiago Moutinho

0 comments