Sensing and Perception

A lone guard stands in a dark hallway, unaware that a thief is creeping silently just inches behind him. How does the game engine decide if that guard should notice the intruder or continue his patrol without any suspicion?

The Mechanics of Virtual Vision

Non-player characters need a way to process their environment to make decisions that feel natural to the player. Most developers use a field of view system to simulate how a human eye captures visual data within a specific cone. This system calculates the distance between the character and a potential target while checking if any physical obstacles block the line of sight. If the target resides inside this geometric cone and nothing obscures the view, the character registers the presence of the player. Think of this process like a security camera that only records when an object enters its active range. Just as a camera ignores movement in a blind spot, the character ignores entities that fall outside its assigned sensory parameters. This creates a fair challenge because players can hide behind objects to avoid detection by the enemy artificial intelligence.

Key term: Field of view — the specific geometric area in front of an entity that determines what it can perceive at any given moment.

Developers must balance these sensory systems to ensure the game remains fun and challenging for the human player. If the vision cone is too wide, the enemy becomes impossible to sneak past, which leads to player frustration. If the cone is too narrow, the enemy feels blind and unintelligent, which ruins the immersion of the experience. The system relies on mathematical checks that run every few frames to keep performance stable during intense gameplay. These checks verify if the target is within the maximum distance and if the angle between the character and the target is within the threshold. By adjusting these values, designers can create different types of enemies with varying levels of alertness. A highly alert guard might have a wide and long vision cone, while a distracted enemy might have a much smaller area of perception.

Processing Sensory Data in Code

To manage these interactions, the game engine uses a structured approach to evaluate incoming sensory data. The following steps outline how the system processes a potential detection event during each frame of the simulation:

1. Calculate the vector between the enemy position and the player position to find the distance.
2. Check if the distance is less than the maximum range defined by the enemy type.
3. Compute the dot product of the enemy look direction and the vector toward the player.
4. Compare the result of the dot product against the vision angle threshold to confirm visibility.
5. Perform a raycast to ensure no walls or solid objects exist between the two entities.

This logic ensures that characters only react when they have a clear line of sight to the player. The raycast acts as a digital laser that verifies if the path is truly clear for the character. Without this final check, characters would be able to see through walls, which would destroy the stealth mechanics of the game. Using these steps allows the AI to react in a way that players can predict and manipulate through clever movement. The combination of distance, angle, and line-of-sight checks provides a robust framework for all sensing behaviors in modern games.

def is_target_visible(enemy, target, max_dist, angle):
    distance = calculate_distance(enemy.pos, target.pos)
    if distance > max_dist: return False
    
    direction = normalize(target.pos - enemy.pos)
    dot_prod = dot(enemy.forward, direction)
    
    if dot_prod < cos(angle / 2): return False
    
    return not raycast_hits_wall(enemy.pos, target.pos)

This code snippet demonstrates the core logic used to verify if a target is visible to an entity. By calculating the dot product, the engine determines if the target is within the field of view angle. The raycast function ensures that physical geometry blocks the sight if the player hides behind a crate. These simple geometric checks form the foundation of how enemies perceive their world in real time. Developers often tweak these functions to create unique enemy behaviors that define the difficulty of the game. Mastering these mechanics allows for complex interactions where players must observe enemy patterns to succeed in their missions.

---

Perception systems enable non-player characters to evaluate their environment by combining geometric vision cones with physical line-of-sight verification.

But what does it look like when these sensory inputs trigger a shift in behavior?