Robust decentralised mean field control in leader following multi-agent systems