A major proposition for collaborative robots is to provide them with enough cognitive intellect so that an ordinary user can communicate with them intuitively, in the most common and casual settings, like, inside a home. For this reason a mobile robot must be provided with enough means to handle multi-modal communication with its human. Intelligent, efficient and natural communication between robots and humans is vital to this aspect of human-robot collaboration. This thesis revolves around adding and selecting additional means of communication, enabling the robots to understand the human commands using multiple modalities while creating intelligent and fluent feedback methods to communicate back to the humans. The focus of this thesis is to explore and test the most natural and efficient way of back and forth communication. By experimenting with few but distinctive modalities and a couple of feedback methods. Humans tend to have a preferred way to communicate or understand the information, also for the robots some implementation or actions are smoother. This thesis also explores the idea by creating the user-study, applying the comparative techniques to generalize these modalities and finding out the favorable one for a specific setting. The evaluation of the results show some preference over some modalities compared to the others depending on the type of task, they also reveal the impact of using a specific modality in particular to using the mobile manipulator platform. The higher level of abstraction can be achieved by running the system at a comfortable pace for a human user and making the interactions more natural and resolving and removing ambiguity during communication. Extensions of this work are discussed finally.