A large part of our analog-circuitry world has historically involved sensor input and output, and it's still the case. Signal conditioning is not only important to system users, it's often tricky and complex from a circuitry perspective due to the inherent and unavoidable nature and idiosyncrasies of the sensors themselves.
Of course, some physical parameters are easier to measure than others. One of the most challenging is torque, which is the product of a force around a pivot point times the length of the associated lever arm. The problem is that, in many cases, the physical member whose torque is of interest is moving over a substantial distance, or worse yet, rotating, so getting the signal from the sensor to the system is difficult. Slip rings, for example, are relatively costly, and also have reliability, cost, and physical-placement problems.
Yet despite the practical difficulties, torque is important parameter in many critical applications, and a window into performance. For example, it's a key indicator of the performance of a bicyclist and his/her actual power output, but it's hard to measure the torque from the pedals and sprocket assembly for obvious reasons. (Measuring cadence under a known load is only an approximate proxy for actual power.)
The obvious solution is to go wireless, but size, weight, and battery life are issues. Fortunately, new technologies are overcoming them: “Wireless Power Meters Help Olympic Athletes,” Design News, July 8, 2008.
What's really interesting to think about–and maybe even dream about–is how new sensor technologies, especially using MEMS-based designs, will allow monolithic or multichip designs which incorporate the sensor, signal conditioning, and a wireless (optical or RF) link. As more sensors go into silicon realizations, for physical parameters such as temperature, light, pressure, force, strain, and fluid flow, there are lots of new opportunities and insights into the world around us and the factors we can't easily observe.
-x x x-