Model driven generative domain engineering (MDGDE) is a domain engineering method aiming to develop optimized, reusable architectures, components and aspects for application engineering. Agents are regarded in MDGDE as special objects having more autonomy, and taking more initiative. Design of the agent involves three levels of activities: logical analysis and design, physical analysis, physical design. This classification corresponds to domain analysis and design, application analysis, and application design. Agent is an important analysis and design tool for MDGDE because it facilitates development of complex distributed system—the mobile robot. According to MDGDE, we designed a distributed communication middleware and a set of event-driven agents, which enables the robot to initiate actions adaptively to the dynamical changes in the environment. This paper describes our approach as well as its motivations and our practice. Model driven generative domain engineering (MDGDE) is a domain engineering method aiming to develop optimized, reusable architectures, components and aspects for application engineering. Agents are considered in MDGDE as special objects having more autonomy, and taking more initiative. Design of the agent involves three levels of activities: logical analysis and design, physical analysis, physical design. This classification corresponds to domain analysis and design, application analysis, and application design. Agent is an important analysis and design tool for MDGDE because it facilitates development of complex distributed system -the mobile robot. According to MDGDE, we designed a distributed communication middleware and a set of event-driven agents, which enables the robot to adaptively to the dynamical changes in the environment. This paper describes our approach as well as its motivations and our practice.