A fundamental part of the implementation of complex robotics systems is the development of control algorithms capable of satisfying multiple hierarchical tasks, guaranteeing motion smoothness in the presence of uncertain environments and unpredicted obstacles. A well-established approach implies the definition of multiple objectives associated with different priority levels, where the major goal is to develop algorithms to handle multiple tasks, smoothly transitioning between different priority levels and avoiding discontinuities. This paper proposes a task coordination approach for wheeled mobile manipulators in human-robot collaboration scenarios. The physical interaction between the operator and the robot arm is measured and translated as a reactive avoidance task for the mobile platform while it follows an arbitrary motion task. The coordination between tasks is managed by an activation algorithm, which implements a virtual capacitance-inspired charging and discharging behavior to ensure smoothness and continuity in the transition between the two tasks. The approach is validated on a Franka Emika Panda Manipulator mounted on an omnidirectional mobile platform.