Abstract: In order to improve the autonomy of robot service, an autonomous ontology-based cognition and planning method for robot service in dynamic home environments is presented. Firstly, the ontology model of the intelligent space is established based on ontology technology. And the user-centered adaptive data-concept conversion mechanism is established after the semantic rules are set up, to integrate the information of intelligent space. Based on these, the inference rule base for service task can be constructed to realize the expansion of the ontology model, and the intelligent space ontology updated in real time is matched with the rule base in order to generate the service sequence of the robot. Finally, the idea of hierarchical task network is used to realize the task planning in JSHOP2 planner. Results of the task planning experiment in intelligent space environment show that the service robot can realize the autonomous cognition in the service task according to the environment and the user's information, and provide personalized services for the user actively. The intelligence level of service can be significantly improved.