Multi-sensor Systems: Diagnostics and Fusion
Generally, sensors are used to acquire relevant information about an environment for purposes of knowledge and control. In this chapter, the “control” aspect refers to a sensor's capacity to respond to specification needs of surveillance and command. The development of this capacity leads to the creation of an acquisition chain in the sensor itself which, although it is an essential element, becomes just one basic element in the overall system.
Actually, beyond obtaining an “image“ or “cartography“ of an environment, resolving the problem requires diagnostics. This can include, for example, a classification of the situations, then a decision leading to an action taken on the environment. Here we refer to the environment in the broadest sense, not only the external elements but also the internal elements of the sensors themselves.
As for the term “diagnostics”, usually rather broadly defined, it refers to classification issues that focus on three important problems:
- finding weak or degraded system modes;
- signal segmentation (or sometimes detecting a specific transient element in a signal);
- signature classification or, in a broader sense, formula recognition.
Different steps appear in the conception and implementation of a diagnostic or control system; they require choices and optimizations associated with evaluation criteria.
Choosing the right sensor is obviously of crucial importance; the quality ...