MINDCROWDNPC & CROWD SIMULATION / UNITY
BILKENT UNIVERSITY · SENIOR DESIGN

Every agent.
Part of a
bigger picture.

MINDCROWD is a behavior-aware NPC and crowd simulation plugin for Unity. It connects navigation, individual NPC decisions, and social behavior in one reusable system.

Explore the project Project documents
CS491/492 · Senior Design Project
MINDCROWD IN MOTION01 / 03
In queues0
Using resources0
Visits finished0

Visitors reserve a place, queue, use a kiosk or rest point, and leave. Close the kiosk to change their destination choices.

Approach → queue → use → leave

Interactive website concept · Unity implementation in development.

01 / THE PROJECTA system of connected decisions

Beyond getting
from A to B.

Believable NPCs do more than follow paths. They make individual decisions, react to events, form groups, and shape the crowd around them.

Our selected features bring together individual NPC behavior, specialized crowd pathfinding, emotion-driven decisions, perception, and group movement in a modular Unity package. Different 3D simulation environments will let us test these capabilities, supported by debug tools and small language models (SLMs).

01

3D simulation environments

Different environments to test the system's capabilities:

  • Evacuation simulation
  • Crowd simulation
  • Medieval War Field
SIMULATION ENVIRONMENTS
02

Resource points & pathfinding

  • Crowd pathfinding with a specialized NavMesh
  • Developer-selectable queue formation at waiting points
  • Templates for points of interest and NPC gathering points
  • Points of avoidance
NAVIGATION & RESOURCES
03

Emotion-driven behavior

  • Emotion templates and character-based behavior
  • Event-based feedback
  • SLM decisions for state transitions
EMOTIONS & DECISIONS
04

Perception

  • Configurable agent field of view
  • Detection of nearby NPCs, objects, and points of interest
  • Awareness of nearby events: fights, hazards, panic, and blocked paths
PERCEPTION & AWARENESS
05

Debug tools & benchmarking

  • Congestion reporting and benchmarking for resource points
  • Logging of utilization, turnaround time, and average time spent
  • SLM congestion detection from logs
DEBUGGING & ANALYSIS
06

Group movement & behavior

  • Making and preserving formations
  • Adapting formations to obstacles
  • Emotion diffusion that influences others in the crowd, especially within groups
FORMATIONS & GROUPS
SELECTED SCOPE

These are the selected features for the project. Implementation details and evaluation results will be documented as development progresses.

THE NAME BEHIND THE PROJECT

MINDCROWD

Behavior-Aware NPC and Crowd Simulation Plugin for Unity

PROJECT NAME PRONUNCIATION

MIND crowd

TEAM ID

To be assigned

PROJECT WEB PAGE

URL to be added after deployment

02 / THE PEOPLEBilkent University

Different perspectives.
One shared project.

A team of five Computer Engineering students exploring the intersection of simulation, behavior, and real-time systems.

ACADEMIC GUIDANCE
03 / PROJECT ARCHIVEThe work, documented

Follow our progress.

CS491 deliverables and project meeting records. Submitted files will be linked as the project develops.

Deliverable list: CS491 course guidelines ↗

Available files open in a new tab. Upcoming items are marked “To be added”.
04 / OPEN DEVELOPMENTFrom idea to implementation

Explore the code.

The plugin and its testing environment have separate homes. Follow the repositories for source code and technical documentation.

SEE IT IN MOTION

Project demonstration

Simulation footage, walkthroughs, and comparative results will appear here.

Demo coming later
05 / WHAT COMES NEXTEvaluation plan

Built to be
put to the test.

We will compare standard NavMesh agents against progressively enabled plugin modules in repeatable scenarios.

Opposing flowsAlternative corridorsCompeting resourcesEvent reactionsGroup movement

Planned measures: travel and waiting time, completion, throughput, congestion duration, and frame time. Results will be shared in the project archive.