Robotics often requires interaction (e.g. supporting a patient in rehabilitation) and therefore a knowledge of how humans perform joint action and a creative way to transfer this to human-robot interaction is needed. The workshop supports this through three aspects: i) joint talks to create cross-discipline discussion; ii) a poster session for field engagement; and iii) open Q&A sessions to spark innovative ideas. Finally, the plenary discussion will analyze the usage of deep learning models for interaction, and explore real-world applications.