Korean Journal of Computational Design and Engineering

Flocking algorithms have been widely applied to simulate group behaviors in virtual environments such as military training, autonomous systems, and game AI. However, traditional flocking models mainly focus on local alignment, cohesion, and separation, lacking the ability to handle dynamic threats or obstacle avoidance in real-time. This study proposes an enhanced flocking algorithm that integrates an avoidance behavior model to simulate more intelligent and strategic movement of virtual enemies. The proposed method incorporates real-time obstacle and threat detection into the flocking process, allowing agents to adaptively adjust their paths while preserving group cohesion. To achieve this, I designed a hybrid algorithm requiring only the leader's path information, thereby reducing the cognitive load on human operators. The algorithm was mathematically modeled and evaluated through scenario-based simulations to assess responsiveness and survivability. Experimental results show that the integrated model improves responsiveness, survivability, and realism of agent behavior in complex environments. These findings indicate strong potential for use in military simulations, swarm robotics, and real-time strategic systems requiring cooperative yet threat-aware autonomous behavior.

키워드 Flocking algorithm, Intelligent agent movement, Obstacle avoidance, Real-time behavior modeling, Virtual enemy simulation