基于软传感器的信息融合

A New Approach for Information Fusion Using Soft Sensor

为了利用"凝视视觉几何约束"的信息来确定移动机器人的位置和方位角,就需要解决数据融合的问题。"几何约束"不是来自真实传感器的直接可测量的数据,这种特殊形式的信息不能被直接融合。为此目的,该文提出了一种融合特殊形式信息的新途径,也即利用"软传感器"的方法来融合来自"几何约束"的信息。软传感器的输出与其它真实传感器的输出一起经过扩展信息滤波器最终实现融合。文中最后提供了利用该方法进行移动机器人定位的计算机仿真例子。仿真结果表明了软传感器信息融合方法的可行性和有效性。软传感器可以广泛应用在很多类似的信息融合问题中。

To determine the position and heading of mobile robot using the information from fixation vision geometric constraint, one information fusion problem must be solved. Geometric constraint is not measurement data from sensor, can not be fused directly. For this aim, this paper presents a new approach to fuse information in special form. The concept of ＂soft sensor＂ is adopted. Outputs of soft sensor and other real sensors＇outputs are both treated by ex- tended information filter for the information fusion purpose. Its application to mobile robot localization is shown with computer simulation in this paper. Simulation results proved the feasibility and validity of soft sensor information fusion. Soft sensor information method can be widely used in similar problems.