Telepresence technology can provide an alternative working style for seniors who want to continuously contribute to society after retirment by delivering their experiences and know-hows to a young generation. Conventional telepresence methods are, however, limited to a face-to-face communication utilizing a static flat-panel display or a physically constrained humanoid robot, which lack the presence of a remote operator, non-verbal communication cues such as a body language, and the freedom of mobility. In this project, we explored a mobile embodied 3D avatar system to shift a rich expressiveness of full-body avatar from virtual world to daily life by integrating a high maneuvability into a human-scale semi-transparent holographic display. We investigated a novel architecture to contorl embodied multimodal avatar by deploying a multimodal behavior generation framework, SAIBA (Situation, Agent, Intention, Behavior, Animation). Synergy effects of a physical movement through a moveable robotic display and a smooth computer graphic animaiton visualized on a 3D display demonstrated a dynamic visual media expression for Spatial Augmented Reality, which can improve the presence and non-verbal communication cues of a remote operator.