How electronic sensors mirror nature
Most people believe humans have five senses: eyes to see, ears to hear, noses to smell, tongues to taste and skin to feel.
In reality, humans have at least nine senses. The sense of thermoception, for example, refers to our ability to perceive temperature. Even blindfolded, if you hold your hand close to something hot or cold, you can perceive the presence or absence of heat.
Some non-human organisms extend the capabilities of the five fundamental senses to an extraordinary degree. Mantis shrimp, for example, have the most complex eye known in the animal kingdom.
Other non-human creatures have additional senses that are different from ours. Bats and dolphins employ forms of echolocation for navigation, foraging, and threat detection and avoidance. Fish can detect movements and pressure changes in the surrounding water using lateral lines of mechanoreceptors. Sharks and some other creatures can detect the electric fields generated by other living creatures using a sense called electroreception.
Early sensors of yesteryear
Early electronic sensors — many of which are with us today — were relatively few, simplistic and limited in capability. For example, magnetic reed switches, which consist of a pair of flexible ferromagnetic contacts (“reeds”) in a hermetically sealed glass tube, can be used to detect when a door is open or closed. Similarly, electromechanical microswitches can be used to detect the presence or absence of an object and are often used to control machinery by defining the limit of travel of a mechanism, which is why these devices are often referred to as limit switches.
More sophisticated sensors were typically characterized by their large size and high cost. In the 1960s, for example, three-axis electromechanical gyroscopic sensors, such as those found in the guidance systems of B-52 strategic bombers, were the size of small oil drums and cost hundreds of thousands of dollars.
How things have changed...
Sophisticated sensors today
The massive growth of the Internet of Things (IoT) in recent years has driven — and been driven by — the development of a cornucopia of sophisticated sensors.
Many of these sensors are presented in the form of microelectromechanical systems (MEMS) that contain both mechanical and electronic components. MEMS are made up of components between 0.001 and 0.1 mm in size, and MEMS devices generally range from 20 micrometers to a millimeter in size. For example, it's now possible to purchase a nine degrees of freedom (9DOF) MEMS sensor that boasts a three-axis accelerometer, a three-axis gyroscope and a three-axis magnetometer — along with a 32-bit Arm Cortex-M0+ processor that performs sensor fusion — all presented in a 4 x 5 x 1 mm package for a couple of dollars.
Although sensors can be used for an incredibly wide variety of applications, it’s hard not to think about them in the context of humanoid robots. For example, some sensors reflect the five fundamental human senses: optical sensors for vision, audio sensors for hearing, olfactory sensors for smelling, gustatory sensors for tasting and tactile sensors to provide a sense of touch. Furthermore, the 9DOF sensor could be used as part of a robot’s proprioception and equilibrioception systems.
What use is all of this? Well, imagine that you are presented with a machine screw and associated nut, and you are told to attach the nut to the screw. You will start by using your eyes to detect the location of the screw and nut and use this information to guide your hands to pick them up. If you start to turn the nut and you feel that it’s requiring too much force, then you may well assume that you’ve got a crossed thread, in which case — rather than simply applying more force and stripping the thread — you will back off and try again. A robot with optical sensors augmented with tactile sensors can adopt a similar strategy.
We can equip robots with senses that we do not ourselves possess, which we might think of as extrasensory perception without involving clairvoyance or telepathy. As one example, a company called Si-Ware Systems offers a small near-infrared (NIR) sensor approximately 10 x 10 mm square that can be used to analyze materials to see what’s there. It won’t be long before you will be able to hold a device (perhaps your smartphone) featuring one of these detectors over a plate of food and be immediately informed of the presence of any of the things to which you are allergic (gluten, nuts, shellfish, etc.).
Another interesting sensor was recently announced by Ultrasense Systems. Smaller than a grain of rice, this sensor — which can be mounted under (or in) a surface — includes a microcontroller, an ultrasonic transceiver that can detect the presence of a finger (for example) and four tiny strain gauges which can detect deformations as small as 100 nm caused by that finger pressing. This sensor is primarily targeted at replacing the common switch, but it could also be embedded in a robot’s hand and used to identify different materials. The speed and characteristics of the ultrasonic signal are modified as a function of the material being addressed by the sensor.
Of course, no matter how sophisticated a sensor is, it’s of limited use in isolation. The real game-changer comes with the combination of these sensors and artificial intelligence, a subset of which is machine learning.
