智能系统多传感器信息融合的复杂性迫切需要开发一套合适的结构体系，目前大多数结构体系都通过融合中心对分散在不同点的多个传感器进行信息处理，而底层传感器之间缺乏必要的联系．这样导致融合中心计算和通信的负担过重而造成瓶颈，且不能使传感器之间互相启发以提高任务环境认知的效率．针对这些问题本文首先提出智能传感器的新概念，指出智能传感器须具备的５个基本能力即预测、规划、刷新、通信和同化，并在此基础上讨论了多智能传感器组成系统时的算法及信息流程．最后以主动视觉和主动触觉共同感知运动物体的位姿为例剖析了这种新思想的具体运用 The complexity of multi-sensor information fusion of intelligent system urgently needs to develop a suitable structural system. At present, most of the structural systems currently process the information of multiple sensors scattered at different points through the fusion center, but the necessary linkages between the underlying sensors are lacking . This causes the fusion center to overcommit and bottleneck the calculation and communication, and can not inspire each other to improve the efficiency of task environment cognition. In response to these problems, this paper first proposed a new concept of smart sensors, pointed out that the smart sensors must have five basic capabilities that are predictive, planning, refresh, communication and assimilation, and on this basis, the algorithm and information Process. At last, we take the active vision and the active tactile sense to jointly perceive the pose of the moving object as an example to analyze the concrete application of this new idea