ConsScale Calculator for FPS Bots

Architectural Level ConsScale Level
(0) Isolated (0) Isolated
CQS exponential curve graph: Cognitive Profile Bar HBar
ConsScale Radar Chart

Line Radar Meter
CLS (0 - 1.55) CQS (0 - 1000) (?)
0 0
Started. Use Reset button to clear all data.
ConsScale Calculator. Raúl Arrabales
Version 3.1(FPS)(en) - Jan. 2016
ConsScale FPS Calculator

ConsScale Calculator for First-Person Shooter Video Game Bots

FPS Bots

See this paper for a detailed explanation of how to rate FPS video game bots using ConsScale.
See below how to rate your bot (or click here for an example).
Click here for a description of ConsScale levels.

ConsScale Calculator for PFS Bots: Input Forms.

1.- Select the architectural components.
2.- Indicate the behavioral profiles (BP).

- Use the checkboxes below to select the features of your FPS Bot.
- The results will appear immediately in the left panel.
- The radar graph might take up to one second to be updated after a change.
- Button [Level 3] will load the profile of a pure ConsScale Adaptive agent.
- Use button [Reset] to start from scratch (instead of page reload).
- Use button [Report] to generate a complete report of your rating.

Name and version of the agent:

Please, enter the name (and version) of your implementation (if you want it to appear when you generate the report):

Agent's Architectural Components:

E First-Person Shooter game environment.
B Software simulated FPS player body.
Spropio Bot has an internal state data acquisition mechanism (Proprioceptive Sensory Machinery).
Sext Bot has a game environment sensory data acquisition mechanism (Exteroceptive Sensory Machinery).
A Bot has a body control mechanism (Action Machinery).
R Bot has a mechanism to generate actions based on sensory information (Sensorimotor Coordination Machinery).
M Bot has a mechanism to store (part of) its past states (Memory).

Mn Bot has capability of multiple context representation in memory.
Att Bot has a mechanism for directing S and A to specific Ei, i.e. Attention mechanism.
SsA Bot has a Self-status assessment mechanism.
I Bot has a mechanism for representation of self.
O Bot has a mechanism for representation of other selves.
AR Bot has an accurate report mechanism.
AVR Bot has an accurate verbal report mechanism.
Rn Bot has a mechanism to run and synchronize several streams of consciousness in one self.

ConsScale levels Behavioral Profiles (BPi,j) for FPS Bots:

Level -1. Disembodied - L-1 = 0.00
N/A No BP considered at this level.

Level 0. Isolated - L0 = 0.00
N/A No BP considered at this level.

Level 1. Decontrolled - L1 = 0.00
N/A No BP considered at this level.

Level 2. Reactive - L2 = 0.00
BP2,1 Basic reflexes: as the ability to back up whenever the bot bumps into another bot or object.

Level 3. Adaptive - L3 = 0.00
Check/Uncheck All.
BP3,1 Ability to learn new simple behaviors adapted to the game: Basic behaviors that help the bot reaching better scores, like shooting other players when they are detected.
BP3,2 Ability to use self state (health, ammo, etc.) to learn new adapted behaviors: Looking for health packs when health level is low or looking for ammo when needed.
BP3,3 Ability to ignore sensory input not critical to current task: Ignoring detected ammo reloading kits when involved in a firefight and no more ammo is needed.
BP3,4 Ability to discard actions not suitable for current situation: Actions like firing to walls when running away from an enemy are considered useless and avoided.
BP3,5 Ability to select what information worth remembering (accessed from memory): When the bot is in need of ammo, it access its memory to get the position of previously seen ammo packs, then it goes directly to pick up the closest one.
BP3,6 Ability to evaluate other players as friends or enemies. Ability to evaluate the benefits obtained by different ammo or health packs: Bot does not attack friends. Healing and re-arming is performed quickly by selecting the best health and ammo packs.
BP3,7 Ability to select what information should be stored in memory: The position of health or ammo packs that could be needed later are stored in memory. The bot goes directly to a remembered position when it needs a pack (see BP3,5).

Level 4. Attentional - L4 = 0.00
Check/Uncheck All
BP4,1 Ability to learn from trial and error: The bot identifies other players as friends or enemies by trial and error. If a player currently considered as friend (see BP3,6) starts attacking the bot, it is now considered as enemy and the corresponding adaptive behaviors are performed (running away or shooting).
BP4,2 Ability to adapt behavior to specific targets: The bot shows directed and sustained behavior towards enemies, like following and shooting them or running away from them.
BP4,3 Ability to evaluate own's performance in combat: Actions that are not contributing to the expected goal are discarded. For instance, running away behavior is changed by another when this behavior is not contributing to diminish damage.
BP4,4 Basic ability to plan next movements: Bot shows a coherent sequence of actions planned in order to reach certain goal. For instance, leaving a firefight for re-arming and then going back to combat.
BP4,5 Ability to keep a depictive representation of objects in the game, i.e. representation in a sensorimotor grounded manner: The bot is able to effectively locate objects and calculate relative positions despite of its changing body and sensor positions (see BP4,2). Bot shows a good shooting accuracy.

Level 5. Executive - L5 = 0.00
Check/Uncheck All
BP5,1 Ability to interleave between game tasks: Behaviors interrupted due to certain circumstances are later resumed (see BP4,9)
BP5,2 Ability to pursue several game goals: For instance, a firefight is eluded because the bot is in need of healing, after getting a health pack, the bot resumes the attack.
BP5,3 Ability to evaluate performance in relation with the accomplishment of several game goals: the bot estimates to what extent goals are being accomplished depending on strategies being used. Effective behaviors are repeated more frequently than behaviors that lead to poor results.
BP5,4 Ability to learn based on game experience: Evaluation performed according to BP5,3 is used to select most promising strategies (see BP5,3). For instance, the bot learns to use most destructive weapons when they are available.
BP5,5 Ability to plan actions taking into account all active game goals: Actions are effectively interleaved as required for the accomplishment of multiple active goals. For instance, trajectory is slightly modified whilst chasing and shooting an enemy in order to pick up some ammo packs available in the surroundings.
CS5,6 Ability to generate and select explicit representations of the world with grounded meaning: the bot autonomously generates internal high-level representations of the events taking place around it, for instance, some percepts indicating a dangerous place in a map are automatically generated by the bot (without being explicitly pre-programmed).

Level 6. Emotional - L6 = 0.00
Check/Uncheck All
BP6,1 Ability to assess global self-status as an actor in the game: This represents functional aspects of emotions. The bot enters a particular state depending on self-status assessment. Therefore, different styles of combat are observed over time (depending on bot's self-status assessment)
BP6,2 Ability to adapt control mechanism to current status: Global behavior is biased by this ("emotional") state. Different behaviors are observed depending on bot's self-assessed state. For instance, it bot's combat skill level is much lower that current enemies' level, its predominant emotional status (e.g. fear) should make it avoid fight instead of chasing enemies
BP6,3 Ability to keep a representation of emotions as described in CS6,1: for instance, if health is very low and no health packs are available, the bot tends to behave as if it was scared, avoiding any risk.
BP6,4 Ability to keep an accurate representation of player's body: The bot control its position, gesture and orientation effectively. For instance, it is able to coordinate its sensorimotor systems to run in one direction while shooting to another relative direction at the same time.
BP6,5 Ability to learn abstract concepts related to the game: Intelligent decisions indicate that specific knowledge about the game has been learnt. For instance, the bot tends to attack lonely enemies and run away from groups of enemies.
CS6,6 Ability to represent a flow of integrated percepts including self-status: the bot autonomously generates a sequence of high-level representations that summarizes current situation. For instance, "I am being attacked and I can't get rid of my enemies".

Level 7. Self-Conscious - L7 = 0.00
Check/Uncheck All
BP7,1 Ability to maintain a model of self and a second order representation of the relation between the self and perceived game action: The behavior of the bot indicates that a sense of self is present. Decisions are not taken just as a function of player state (health, ammo, etc.), but based on a rich model of self which constitutes the basis for Theory of Mind capabilities.
BP7,2 Ability to maintain an analogous second order representation of the relation between the self and bot actions: The bot is able to recognize itself and the consequences of its own actions. In other words, a sense of agency is developed and it can be perceived in bot's behavior (see below).
BP7,3 Ability to maintain a second order representation of the relation between feelings and self: actions are also evaluated in terms of their consequences for bot's self status.
BP7,4 Ability to self-recognize as a player in the game: Possible behavior tests include mirror test derivates.
BP7,5 Ability to make plans including the model of self as an actor: Behavior is modulated by the ability of the bot to foresee (remember) the emotional outcome of a planned known action.
BP7,6 Ability to imagine the outcome of planned actions in terms of self: Behavior is also modulated by the ability of the bot to foresee (imagine or internally simulate) the outcome of planned new actions. Therefore, new behaviors appear as a result of advance planning mechanism including imagination. For instance, the bot develop new strategies to attack enemies that have not been learned using reinforcement but imagination.
BP7,7 Ability to use existing game objects as tools (note that support for using weapons and vehicles is native in the game, so their usage cannot be regarded as a bot cognitive capability): The bot manages to use some object as a means to achieve its objectives. For instance, using a movable object, like a box or a barrel, as an improvised shield.
CS7,8 Ability to represent and self-report mental content (continuous inner flow of percepts): The bot internally plays a "movie" as a sequence of percepts representing the events taking place in the surroundings, including own body actions. That movie is also used as input for decision making.

Level 8. Empathic - L8 = 0.00
Check/Uncheck All
BP8,1 Ability to model other players as intentional selves: As other players are identified and modeled as selves, their movements can be predicted. The bot put itself in the place of another player to predict next actions of an opponent. Then, the behavior of the bot is shaped not only according to present sensory data but also using oponent's predicted movements. For instance, the bot predicts the possible escape path of an enemy and make the necessary moves to block it.
BP8,2 Ability to learn from other players by imitation: As the bot can manage both the model of self and models of others, it can also establish analogies and learn strategies by observing other bots. For instance, the bot can acquire new attack strategies developed by human players participating in the same game.
BP8,3 Ability to collaborate with other players to get better scores: Social behaviors like forming groups that collaborate in firefights.
BP8,4 Ability to make plans including the models of other players as actors in the plans (intersubjectivity): Social behaviors like forming groups that collaborate in firefights.
BP8,5 Ability to build new tools than can be used to achieve game goals: The bot combines several objects in order to build a new compound object that can be used either for defense or attack. For instance, building an improvised barricade made of a number of objects arranged along a line.
CS8,6 Inner imagery is enriched with mental content related to the model of others: the bot keeps an updated models of other players and use these models to build a richer inner imagery representing current world state. For instance, "I'm attacking player XYZ, who is usually scared of me".

Level 9. Social - L9 = 0.00
Check/Uncheck All
BP9,1 Ability to develop Machiavellian as part of the game play: The bot is able to reason about opponents' Theory of Mind, i.e. "I know you know I know". Therefore, it shows social intelligent behaviors like preparing an ambuscade.
BP9,2 Learning of new Machiavellian strategies: The bot learns new Machiavellian strategies that have not been pre-programmed, for instance new ways of deceiving enemies.
BP9,3 Ability to report mental content: The bot uses game's inbuilt chat system to coherently report its inner mental state.
BP9,4 Ability to form cultural groups: Behavioral profiles associated with culture would require more complex environments. However, clues of cultural organization might be observed in groups of organized bots.
BP9,5 Ability to modify the environment to serve bot's needs: Like in BP9,4 complex behaviors associated with these skills requires more complex environments. Nevertheless, bot autonomously using movable objects of the map to build a secured base would be a hallmark of this skill.

Level 10. Human-Like - L10 = 0.00
Check/Uncheck All
BP10,1 Ability to produce accurate verbal report: The bot generates complex linguistic messages using correct grammar and accurate meaning.
BP10,2 Ability to pass an FPS adapted version of Turing test: The bot will pass an adapted Turing test, like the one proposed in the BotPrize competition. Also classical Turing tests using game chat could be passed.
BP10,3 Ability to develop civilizations and technology: Like in BP9,4 complex behaviors associated with these skills requires more complex environments.

Level 11. Super-Conscious - L11 = 0.00
BP11,1 Ability to manage several streams of consciousness: For instace, the bot is able to keep a complex explicit conversations over the chat with human beings (all passing the Turing test), and fight enemies effectively at the very same time. All conscious threads share knowledge, i.e. parallel chats could comment on the ongoing deathmap game