Over decades, sensors were high-precision devices but their function was simple and one-dimensional. And they were analog. Nowadays, the world no longer is that simple " sensors get digital and their properties are defined by their algorithms. This creates a new complexity that calls for methods to handle.
Sensors this are these little devices that transform non-electrical physical units into voltages or currents, right? For instance, pressure in voltage or, in less straightforward cases, resistance in frequency or so. Simple. Pressure (or temperature or whatever) in, voltage out. Not much more difficult to handle than a resistor.
But even the simple sensor world is growing more complex. Not because an increasing percentage of sensors deliver their data not only or not at all in analog form but via a digital interface. No, this would be easy to handle; after all, it is just one complexity level up. What makes the world of sensors complex and interesting is that nowadays they start to combine several different physical units and combine these values by means of algorithms. Microprocessors and digital signal processors increasingly become integral part of the sensor landscape.
At the Sensor + Test trade fair in Nuremberg, innovators displayed their achievements, hinting at future developments. In their sensors to measure lubricant oil degradation, even vendors of relatively simple devices combine several MEMS with signal digitalization circuitry, signal preprocessing, calibration and temperature compensation. All digital, of course.
An more complex example showed the Fraunhofer Technology Development Group (Fraunhofer TEG) from Stuttgart. They combined low-cost three-axis inertial MEMS platform with a camera that translates visual patterns into spatial information, creating a highly precise navigation system which is completely independent of GPS or other external signals and thus suited for indoor use. All digital and, unnecessary to mention, the actual intelligence was not in the silicon nor in other parts of the hardware but in the algorithms.
Some people in this situation ask "are these still the sensors we used to use?", but I think this is the wrong question. The more interesting topic is that the applications possible with this class of sensors are much more complex than we could create with simple analog pressure or temperature sensors, and we have to find ways to handle this complexity. This question still waits to be resolved.
And, by the way, the simple analog sensors don't go away. They will continue to be around since simple solutions for simple problems will need to be resolved also in the future.