<p>Upper-limb exoskeleton (ULE) arms have the potential to assist humans in accomplishing tasks by distributing a heavy load. ULE arms are designed to be comfortable and lightweight wearable robotic devices, a design that has resulted in the complexity of their structures, actuators, and power transmissions. Additionally, different ULE vendor structures and biomechanical variations between humans have resulted in dissimilar coordinated ULE arm systems. These complex multiple ULE arm systems can be handled through adaptive virtual decomposition control (VDC) if unknown dynamic models are not considered. Accordingly, this paper proposes a new distributed framework for the adaptive impedance-based VDC method to address the abovementioned challenges and thereby enhance the performance and robustness of the ULE arm systems. To that end, the proposed control method has a prediction capability and an ancillary control law for coordinated dissimilar ULE arms holding a common object. The system stability is analyzed using the input-to-state stability approach. The performance of the proposed controller is evaluated both in simulation with six coordinated ULE arms and in an experiment with two commercial coordinated ULE arms each with seven degrees of freedom. Four scenarios are performed with different internal arm forces with and without an obstacle. The simulation and experiment results are compared with a state-of-the-art adaptive VDC method and show the superiority of the proposed control method.</p>