It is theoretically argued that multi-technology fusion has a life cycle like any product innovations. It begins as a researcher starts socialization in doing a research work to integrate a different technological knowledge. Then, an innovation stage emerges as the researcher generates a new knowledge that creates new functions or new product. Innovative knowledge become a routine knowledge as the researcher stops to add a new things moving to a stabilization stage that finishes a life cycle of multi-technology fusion. Innovation of intelligent robotics has a system nature with multi-technology fusion because various professionals and institutes play roles in creating new technology. It starts as a researcher with a cognitive map on the one technology in a firm does interact with another researcher holding another cognitive map on a university or a research institute. Their interaction evolves to collective learning with collective cognitive maps. This collective learning is distinctive feature of robotics innovation by multi-technology fusion in which core element is knowledge activities such as acquisition, assimilation, integration and creation of different robotics technology for a new robot. This paper aims to understand the mechanism and patterns of R&D activities among different organizations like universities, research institutes and firms for an innovation of robotics technology through multi-technology fusion. It investigates the intelligent robotics industry in both Korean and Japan by analyzing a meso-level statistics and taking interviews with professional people.