Efficient Multi-Robot Cooperative Exploration for Semi-Enclosed Environments Based on Multi-Constraint Consensus
Multi-robot systems enable rapid and scalable environmental exploration by distributing sensing and coverage across multiple robots. However, in semi-enclosed environments with complex connectivity (e.g., office, hospital), existing methods fail to account for local reachability and topological structure, leading to inefficient exploration. In this letter, we propose a cooperative exploration framework for such semi-enclosed environments. The framework first partitions the global mission into spatially continuous task domains using spatial correlation-aware domain fusion (SCADF). These task domains integrate topological relationships and local navigability and align the task geometry with the environment. Task assignment is then performed by a novel multi-constraint consensus-based bundle algorithm (MC-CBBA). MC-CBBA augments the classical CBBA with three constraints: inter-robot workload balance, minimum per-robot workload guarantee, and path cost minimization. A two-stage distributed consensus protocol ensures feasible and balanced allocations. Simulation experiments indicate that the proposed method reduces mission completion time by an average of 36% compared with state-of-the-art methods, while significantly enhancing load balancing and exploration efficiency. Real-world experimental results further demonstrate the robustness and effectiveness of the proposed approach.