Cooperative Control of Multi-Agent Systems: A Consensus Region Approach