Abstract: Cooperative obstacle avoidance and formation control in complex environments remain key challenges in multi-mobile robotic systems. Traditional centralized control methods are constrained by single-point failures and computational bottlenecks, whereas distributed methods often exhibit limitations in maintaining formation stability. To address these issues, a distributed control framework is proposed. In this framework, a finite state machine is integrated with the Raft consensus algorithm to enable robust cooperative obstacle avoidance and formation maintenance in complex environments. The framework employs a quantified conflict metric to drive state transitions. When conflicts intensify, a decentralized leader re-election is activated through the Raft consensus algorithm. The newly elected leader synchronizes reference trajectories via log broadcasting, while the followers perform trajectory optimization by incorporating local perception information. In a formation consisting of seven robots arranged in a regular hexagonal pattern, the proposed method achieves an average path length of 319.877 m and an average task completion time of 123.65 s in dense obstacle environments, while demonstrating superior formation stability compared to typical distributed approaches. Simulation results validate the effectiveness of the proposed method in accomplishing the objectives of multi-robot cooperative collision avoidance and formation control in complex environments.