Despite the recent developments in robotic technology that have introduced robots in various fields of industry, agriculture, service, security, etc., complex applications that involve multiple robots performing a task in physical interaction with humans, require further investigation. Towards this direction, we propose an impedance control scheme for multiple mobile manipulators assisting a human in transporting/manipulating a rigidly grasped object. The human operator plans the object's motion, whereas the robot-team imposes predefined impedance properties on the object to reduce the human effort required to execute the task, without assuming either knowledge of the desired object trajectory or explicit inter-robot communication. Moreover, to achieve the desired impedance properties two novel decentralized estimation schemes are employed to derive the human motion intention and the human exerted wrench. Finally, a simulated paradigm verifies the theoretical findings.