This paper presents a dynamic leader-follower formation control mechanism that steers a swarm of flying drones through narrow spaces in complex environments. Many approaches have been used for formation control in swarm robotics; however, the bio-inspired Optimised Collective Motion (OCM) model that is used in this study minimises fluctuations among the robots and improves swarm coordination. The proposed control mechanism enables dynamic reconfiguration, allowing the swarm to switch between square and line configurations to adapt to environmental constraints and dynamics. Extensive simulations using the Monte Carlo approach demonstrate the robustness of the proposed algorithm. Real-flight experiments with Crazyflie drones confirm the feasibility of the control mechanism in achieving stable flocking behaviour in cluttered environments.