<p>This paper proposes an environment-adaptive formation planning and control framework for multi-robot systems operating in confined indoor spaces. In such environments, robots must navigate while avoiding collisions with walls, doorways, and other robots. The multi-robot exploration of confined spaces requires the generation of command formation patterns and real-time adaptation of these patterns to ensure collision-free movement. The proposed algorithm dynamically selects collision-free formations from a predefined set of formation patterns based on a local two-dimensional occupancy grid map. This allows the robot group to adaptively adjust its robot formation in response to environmental changes while maintaining safe distances from obstacles. For robot control, the proposed formation planning algorithm is integrated with a model predictive control (MPC) framework that computes individual control commands that guide each robot along the selected formation while avoiding collisions with obstacles and other robots. The proposed algorithm was validated through a series of virtual experiments conducted in a Gazebo simulation environment integrated with a robot operating system (ROS) to reflect real-world operational conditions. The results demonstrate that the proposed method allows multi-robot systems to adapt to previously unmapped corridors or narrow passages while maintaining robot formation and successfully navigating through narrow and confined spaces.</p>