Considerations

There are 3 kinds of considerations:

• Boolean consideration: Returns a score of 0.0 or 1.0.
  • Use this if you need to consider a Yes-No question. If the answer is yes, it returns 1.0, and vice versa.
  • For example:
    • IsTargetInAttackRange
    • IsTargetInDamageArea
    • IsInAttackCooldown
• Variable consideration: Returns a score from 0.0 to 0.1.
  • Use this if you need a consideration that returns a score that changes dynamically based on the current input.
  • For example:
    • TargetInSightRadius
    • TargetInAttackRange
    • MyHealthIsLow
    • TargetHealthIsHigh
• Constant consideration: Returns a constant score in [0.0, 1.0]. Use this when:
  • You need a consideration that always returns a constant score, e.g., 0.1 or 0.2, etc.
  • You need a fallback decision that will be selected if the the agent doesn’t know which decision to choose in the current situation.
  • For example:
    • Idle

Common Consideration Recipes¶

Distance¶

IsTarget(Not)InRange¶

• Returns 1.0 if the target is within the specified range or 0.0 if it is outside the range.
• Recipe
  • Input: DistanceToTargetInput
  • InputNormalization: IsInRangeNormalization
  • ResponseCurve: Basic Linear (Inverse Linear)

TargetInRange¶

• Returns a score in [0.0, 1.0]. It maps the input value (DistanceToTargetInput) from [Start, End] to [0.0, 1.0].
  • If the input value is less than Start, returns 0.0
  • If the input value is greater than End, returns 1.0
• Recipe
  • Input: DistanceToTargetInput
  • InputNormalization: InRangeNormalization
  • ResponseCurve: Which ResponseCurve you should use

Is(StateName)State¶

• Returns 1.0 if input state is the specified state; otherwise returns 0.0.
• Recipe:
  • Input: User Custom Input (often an Enum)
    • Returns a state of the agent or the target.
  • InputNormalization: User Custom Input
    • Returns 1.0 if input state is the specified state.
    • Returns 1.0 if input state is not the specified state.
  • ResponseCurve: Basic Linear (Inverse Linear)

Is(Not)InCooldown¶

• Returns 1.0 if the CooldownElapsedTimeInput is within the cooldown duration; otherwise returns 0.0
• Recipe:
  • Input: CooldownElapsedTimeInput
  • InputNormalization: IsInCooldownNormalization
  • ResponseCurve: Basic Linear (Inverse Linear)

Random¶

• Returns a random score in [0.0, 1.0]
• Recipe:
  • Input: User Custom Input
    • Returns a random input value.
  • InputNormalization: BasicNormalization
  • ResponseCurve: Basic Linear

Health¶

• The input value will be normalized by dividing by 100
  • If the input is less than 0.0, returns 0.0
  • If the input is greater than 100, returns 1.0
• Recipe
  • Input: Health
  • InputNormalization: DivideByMaxValue
  • ResponseCurve: Which ResponseCurve you should use

Idle¶

• Returns a constant score (often 0.1)
• Recipe:
  • Input: None
  • InputNormalization: None
  • ResponseCurve: Constant

Which ResponseCurve you should use¶

• Boolean considerations:
  • Basic Linear or Inverse Linear
• Variable considerations:
  Suppose the input gradually increases from 0.0 to 1.0:
  • The score is proportional to the input and increases gradually.
    • Linear: Basic Linear
    • Slow at first, fast later: Basic Quadric Lower Right
    • Fast at first, slow later: Basic Quadric Upper Left
    • Slow at either end, fast in the middle: Basic Logistic
    • Fast at either end, slow in the middle: Basic Logit
  • The score is inversely proportional to the input and decreases gradually.
    • Linear: Inverse Linear
    • Slow at first, fast later: Basic Quadric Upper Right
    • Fast at first, slow later: Basic Quadric Lower Left
    • Slow at either end, fast in the middle: Inverse Logistic
    • Fast at either end, slow in the middle: Inverse Logit
  • The score fluctuates
    • Slow at either end, fast in the middle: Basic Bell Curve or Inverse Bell Curve
    • Fast at either end, slow in the middle: Basic Logit or Inverse Logit
    • More dynamic: Basic Sine or Inverse Sine
• Constant considerations:
  • Constant

If you like Utility Intelligence, please consider supporting it by leaving a 5-star review on the Asset Store. Your positive feedback motivates me to keep improving and delivering more updates for this framework.
Thank you so much for your support. I love you all! 🥰