Fullscreen HTML5Med Res (??MB)HTML4

That makes so much sense! Techno Talk Max the Magnificent I’m always amazed by the cunning creations of the pioneers of yesteryear, especially when I consider the rudimentary sensors they had at their disposal. I often wonder what their reactions would be to see the sophisticated sensors we have available to us today. A s usual, my poor old noggin is full of random thoughts bouncing around like super balls on steroids. The topic that has currently captured my attention is that of sensors. What pops into your mind when you hear (or see) the word ‘sensor’? Living in an age of wonders as we do, you may be thinking of highfalutin’ devices like lidar (laser detection and ranging) or radar (radio detecting and ranging) . When we boil things down, however, a sensor is any device that detects some physical phenomenon and produces a corresponding output signal. For our purposes here, we will assume electrical output signals in terms of voltage or current being used to feed electrical or electronic systems, but this isn’t cast in stone. Victorian fax machines Can you imagine the Victorians sending faxes to each other? This may seem farfetched, but in 1842, a Scottish engineer and inventor called Alexander Bain came up with a cunning idea. He created an image to be transmitted by cutting it out of a thin sheet of tin. He placed this metal image on a movable insulated base and connected it to one side of a battery. The base was slowly passed under a swinging pendulum formed from a conducting wire with a weighted point on the end. Whenever this point connected with the metal image, it completed the electrical circuit, thereby converting the dark and light areas of the image – which were represented by the presence or absence of tin – into an electrical signal. This cunning creator used this electrical signal to activate a relay attached to the end of another pendulum that was swinging back and forth over a second moving bed. The activated relay caused an attached pencil to encounter a piece of paper laying on the moving bed, thereby reproducing the original image in metal as a drawing in pencil. Ray guns and TV controllers Did you ever see the original Buck Rogers science fiction serial from 1939 starring Buster Crabbe? Filmed in glorious blackand-white, this was originally released as 8 a series of 20-minute movies, and later shown in the 1950s and 1960s on TV. Buck sported a ray gun in the form of a ‘U-235 Atomic Pistol’. The reason I mention all this is that the first practical photoelectric cells were invented in the 1880s. In 1955, Zenith introduced the world’s first wireless television remote control called the Flash-Matic. Looking like something Buck Rogers would not be ashamed to be seen carrying, this glorified torch (flashlight) employed a beam of light to activate four photocells located at the corners of the screen, thereby allowing the user to control the volume and channel selection. Mobile sensor platforms Do you remember the artificial intelligence (AI) called KITT (Knight Industries Two Thousand) powering the highly advanced, very mobile, robotic automobile in the Knight Rider TV series of the 1980s? Today’s cars are getting close to (sometimes they surpass) KITT’s capabilities. My own 2019 Subaru Crosstrek is equipped with binocular vision that can be used to detect and correct any drifting out of lane, vary the speed of the cruise control if we get too close to a car in front, and slam on the breaks if it feels we are in danger of imminent collision. All of this is made even more exciting by my wife screaming in my ear. In fact, today’s autonomous cars and robots are essentially mobile sensor and computing platforms. A very common scenario is to have multiple cameras equipped with CMOS sensor arrays that are sensitive to light in the visible part of the spectrum. These feed advanced processors running AI algorithms that can perform tasks like object detection and recognition. These cameras can be augmented by lidar and radar sensors. The original lidars were big, bulky, and expensive, but new versions are coming online in which almost everything is implemented in semiconductor form. As opposed to a simple time-of-flight (TOF) approach which involves generating powerful pulses of light and measuring the round-trip time of any reflections, companies like SiLC Technologies are using a frequency modulated continuous wave (FMCW) approach that can provide distance and velocity data on a pixel-by-pixel basis, allowing them to perceive and identify objects more than a kilometer away. Meanwhile, companies like Owl Autonomous Imaging are creating longwave infrared (LWIR) thermal focal plane arrays (thermal imagers). The signals from these images can be employed by AI to perform object detection, classification and ranging. As the folks at Owl told me, ‘Within five years, all new cars will be able to see at night!’ That’s deep! Have you ever thought about our amazing ability to perceive the world around us in three dimensions? Powered by our optical sensors (eyes) and associated computers (brains), we call this ability ‘depth perception.’ There are many aspects to this, but we start with the fact that each of our eyes sees a slightly different image due to their separation in our heads. The resulting disparities are processed in the visual cortex of our brains to yield depth information. Even with one eye closed, we can still do things like track and catch a ball heading our way. In this case, our brains make use of visual cues, including knowing how big we expect objects to be, and our understanding that if an object appears to be growing bigger, then this may be a good time to duck. In the case of machine vision, one of the components of depth perception is the ability to create a 3D depth map (point cloud) of the scene. We can do this using two CMOS sensors to provide binocular vision, but that increases the cost. We can employ a single CMOS sensor in conjunction with an AI, using its understanding of the scene to determine where and how big things are in 3D space, but this requires a lot of computation. The folks at a company called AIRY3D have come up with a way to use a single CMOS sensor to generate both a regular 2D image and a 3D point cloud on a pixel-by-pixel basis with very little computation. I don’t know about you, but I certainly didn’t see this one coming! Practical Electronics | August | 2024