课题面向家庭应用背景，针对混合云架构下的服务机器人，模拟人类认知机制和心智发育机理，综合利用实时情景信息与云端知识，通过语义知识推理实现云机器人对服务任务的主动认知，以提高云机器人的智能化水平和个性化服务能力。.课题主要研究内容包括：（1）基于智能空间实现机器人对服务环境和服务对象的泛在感知，并通过将视觉选择性注意机制与关键点-关键区域信息相结合，选择空间感知注意焦点，实现对服务任务敏感信息的高效主动获取。（2）研究感知信息的知识表征方法，并构建与混合云架构相容的、带有深度结构的、覆盖低层个性化知识到高层共性化知识的类脑层次化记忆体系。（3）结合云端知识和情景上下文进行人体行为理解，建立规则权重可自适应调整的、基于语义的服务推理机制，实现服务任务自主推理认知。（4）设计基于层次化迁移学习的知识增长机制，实现云机器人智能水平不断提升。

Considering of hybrid cloud framework for home service robot, the purposes of this project are to simulate the mechanism of human cognition and mental development, utilize the real-time scene information and knowledge in cloud to realize the active cognition to service task by the semantic knowledge reasoning and improve the cloud robot’s intelligence and the ability of personalized services..The main research work of this project includes the following parts: (1) robot’s ubiquitous perception to service environment and human is realized with the support of intelligent space, the visual selective attention mechanism combined with key point-key area information is used to choose the focus of attention and realize the autonomous information acquisition oriented to service sensitive area efficiently. (2) study the knowledgeable expression method of the information, construct the brain-like hierarchical memory system which is compatible with the hybrid cloud, have deep structure and cover the knowledge from personalized knowledge on low level to common knowledge on top level. (3) recognize the human behavior using the cloud knowledge and scene context, establish the semantic knowledge reasoning mechanism in which the weight can be adjusted adaptively and complete the active cognition to service task. (4) the knowledge growth mechanism based on hierarchical transfer learning is built to improve the intelligent ability of cloud robot.