Wavelet-Based Fault-Tolerant Integration and Target Recognition inMultidimensional Sensor Signal Processing

摘要

Sensor fusion research has been motivated by the difficulty of implementingautomated systems which interface with the real word. For a system to react properly to a changing environment, the system requires sensor input to model its environment. Unfortunately, sensors are electromechanical devices in most cases and subject to physical limitations. There physical limitations manifest themselves as constraints on the accuracy, precision, and dependability of the data returned by the sensors. It is computationally challenging for an autonomous system to robustly evaluate sensor data of questionable accuracy and dependability. Sensor fusion seeks to solve this problem by taking inputs from several physical sensors and merging the individual physical sensor readings into a logical sensor reading. This has several advantages, some of which are particularly salient to this research effort. The use of heterogenous physical sensor allows a logical sensor to be developed which is less sensitive to the limitations of any single sensor technology.