WEBVTT 00:00:00.000 --> 00:00:02.459 Picture the camera mounted on the highway gantry 00:00:02.459 --> 00:00:04.320 you drove under this morning. Right. Or maybe 00:00:04.320 --> 00:00:06.540 it was one of those sleek little gray boxes perched 00:00:06.540 --> 00:00:08.939 on a stuck metal pole by the interstate. Yeah, 00:00:08.939 --> 00:00:11.039 exactly. And you probably tapped your brakes, 00:00:11.119 --> 00:00:13.179 assuming it was actively shooting a radar beam 00:00:13.179 --> 00:00:15.619 at your car to clock your speed. Which is what 00:00:15.619 --> 00:00:18.219 we all do, right? Good for sure. Yeah. But what 00:00:18.219 --> 00:00:20.199 if it wasn't emitting anything at all? What if 00:00:20.199 --> 00:00:23.280 it was just watching you using pure geometry 00:00:23.280 --> 00:00:25.559 to calculate your velocity? That's a pretty wild 00:00:25.559 --> 00:00:29.420 thought. It is. Welcome to today's Deep Dive. 00:00:29.579 --> 00:00:33.280 We have an incredibly revealing excerpt from 00:00:33.280 --> 00:00:36.500 a Wikipedia article detailing a technology called 00:00:36.500 --> 00:00:41.829 video detection and ranging or Vidar. And our 00:00:41.829 --> 00:00:44.649 mission today is to decode how modern speed enforcement 00:00:44.649 --> 00:00:47.670 has quietly shifted from bouncing active signals 00:00:47.670 --> 00:00:50.369 off your bumper to passively calculating your 00:00:50.369 --> 00:00:53.829 speed using advanced stereoscopic math. Okay, 00:00:53.969 --> 00:00:56.030 let's unpack this. Well, to really appreciate 00:00:56.030 --> 00:00:58.350 the sheer leap in engineering that Vidar represents, 00:00:58.530 --> 00:01:00.750 we kind of have to look at what the source material 00:01:00.750 --> 00:01:03.130 identifies as the old guard. Right, the traditional 00:01:03.130 --> 00:01:07.019 methods it's replacing. Exactly. The text specifically 00:01:07.019 --> 00:01:09.140 points to two foundational technologies. You 00:01:09.140 --> 00:01:12.900 have radar, which relies on Doppler shifts, and 00:01:12.900 --> 00:01:15.819 LiR, which calculates speed using the time of 00:01:15.819 --> 00:01:18.150 flight principle. OK, we all know the basics 00:01:18.150 --> 00:01:20.790 of those old school systems, but let's translate 00:01:20.790 --> 00:01:22.510 that for the listener really quick. Yeah, good 00:01:22.510 --> 00:01:26.269 idea. So radar is reading the Dopple shift, which 00:01:26.269 --> 00:01:28.909 is basically measuring how compressed a radio 00:01:28.909 --> 00:01:31.629 wave gets when it bounces off the grill of a 00:01:31.629 --> 00:01:34.269 car moving toward the sensor. Right, the frequency 00:01:34.269 --> 00:01:36.689 change. Yeah, it's like driving a boat into oncoming 00:01:36.689 --> 00:01:39.450 waves. The faster you drive, the faster the waves 00:01:39.450 --> 00:01:41.810 hit the bow. So the radar gun measures that wave 00:01:41.810 --> 00:01:44.469 compression to find your speed. Spot on. And 00:01:44.469 --> 00:01:46.989 then there is Liar, which abandons the Doppler 00:01:46.989 --> 00:01:49.109 effect entirely. Because it uses light instead 00:01:49.109 --> 00:01:52.209 of radio waves, right? Exactly. As the text outlines, 00:01:52.689 --> 00:01:55.030 LIDAR relies on time of flight. So think of time 00:01:55.030 --> 00:01:56.909 of flight like bouncing a tennis ball against 00:01:56.909 --> 00:01:59.129 a brick wall in the dark. Oh, I like that analogy. 00:01:59.250 --> 00:02:01.709 You throw it, and you count the exact milliseconds 00:02:01.709 --> 00:02:03.650 it takes to return to your hand to figure out 00:02:03.650 --> 00:02:06.709 how far away the wall is. Just way faster. Right. 00:02:06.909 --> 00:02:10.210 An incredibly fast game of Marco Polo played 00:02:10.210 --> 00:02:13.620 with light instead of sound. The police scanner 00:02:13.620 --> 00:02:16.560 yells, Marco, by firing a laser pulse millions 00:02:16.560 --> 00:02:20.419 of times a second, and it tracks the exact microsecond 00:02:20.419 --> 00:02:22.840 your bumper yells, Polo, back by reflecting that 00:02:22.840 --> 00:02:24.979 light. And since we know the speed of light is 00:02:24.979 --> 00:02:27.860 a constant, measuring the round -trip time of 00:02:27.860 --> 00:02:30.840 that laser pulse gives you the exact distance 00:02:30.840 --> 00:02:33.259 to the car. And you just do that thousands of 00:02:33.259 --> 00:02:35.379 times a second, and boom, you have the car's 00:02:35.379 --> 00:02:38.240 speed. Exactly. But what's fascinating here is 00:02:38.240 --> 00:02:40.879 the underlying operational philosophy of both 00:02:40.879 --> 00:02:43.969 radar and lidar. What do you mean by philosophy? 00:02:44.469 --> 00:02:47.169 Well, despite using totally different physics, 00:02:47.389 --> 00:02:49.669 you know, compression of radio frequencies versus 00:02:49.669 --> 00:02:53.050 the speed of light, they share one defining characteristic, 00:02:53.349 --> 00:02:55.250 which is they're both active systems. Meaning 00:02:55.250 --> 00:02:57.610 they have to like throw energy into the environment 00:02:57.610 --> 00:03:00.490 to get a result. Yes. They are completely reliant 00:03:00.490 --> 00:03:03.599 on a mission to measure the physical world. they 00:03:03.599 --> 00:03:05.400 have to physically interact with it. Oh, I see. 00:03:05.599 --> 00:03:08.400 Like, a LIDAR system requires a laser diode to 00:03:08.400 --> 00:03:11.460 generate intense pulses of light, optics to focus 00:03:11.460 --> 00:03:14.759 that beam, and often some sort of mechanical 00:03:14.759 --> 00:03:17.680 or solid -state scanning mechanism to sweep that 00:03:17.680 --> 00:03:20.139 beam across the highway. That sounds complicated. 00:03:20.360 --> 00:03:23.599 It is. And radar requires an oscillator to generate 00:03:23.599 --> 00:03:26.419 microwaves, an antenna to broadcast them, and 00:03:26.419 --> 00:03:28.819 a receiver sensitive enough to catch the scattered 00:03:28.819 --> 00:03:31.340 reflection. Which means they require significant 00:03:31.340 --> 00:03:34.120 power. Yes, a lot of power. And they have moving 00:03:34.120 --> 00:03:37.080 parts or complex broadcasting hardware that can 00:03:37.080 --> 00:03:39.520 degrade over time. Plus, they're constantly shouting 00:03:39.520 --> 00:03:42.180 into the electromagnetic spectrum. Exactly. And 00:03:42.180 --> 00:03:44.599 that active emission is exactly the paradigm 00:03:44.599 --> 00:03:47.400 our source material says is ending. Wow, really? 00:03:47.759 --> 00:03:50.800 Yeah. We are moving from a regime of active interrogation, 00:03:51.240 --> 00:03:53.479 like pinging vehicles with energy, to a regime 00:03:53.479 --> 00:03:55.939 of entirely passive observation. Which brings 00:03:55.939 --> 00:03:59.099 us to VEDAR. Right. We are entering the era of 00:03:59.099 --> 00:04:01.900 VDAR. So the text defines video detection ranging 00:04:01.900 --> 00:04:04.259 as a technique that measures the speed of a distant 00:04:04.259 --> 00:04:06.580 vehicle simply by tracking the object through 00:04:06.580 --> 00:04:09.099 vision cameras. Just watching it. No lasers, 00:04:09.479 --> 00:04:12.259 no radio waves, just capturing the ambient light 00:04:12.259 --> 00:04:14.560 that is already bouncing off your car. It is 00:04:14.560 --> 00:04:17.079 an incredibly elegant concept. You strip away 00:04:17.079 --> 00:04:19.040 the emitters, the diodes, the antennas, and you 00:04:19.040 --> 00:04:21.220 just replace them with a lens and a sensor. I 00:04:21.220 --> 00:04:23.139 mean, I'm hung up on the mechanics here, though. 00:04:23.259 --> 00:04:26.319 How so? Well, a camera captures flat two -dimensional 00:04:26.319 --> 00:04:28.459 images, right? Right. And a car is moving toward 00:04:28.459 --> 00:04:31.660 the lens in three -dimensional space at like 00:04:31.660 --> 00:04:35.120 75 miles an hour. Yeah, fast. So how does a flat 00:04:35.120 --> 00:04:38.720 grid of pixels calculate velocity without a radar 00:04:38.720 --> 00:04:41.040 wave telling it the physical distance? I mean, 00:04:41.100 --> 00:04:43.339 a single video is just a rapid sequence of flat 00:04:43.339 --> 00:04:46.500 photographs. That exact physical limitation is 00:04:46.500 --> 00:04:49.519 what kept passive video speed detection in the 00:04:49.519 --> 00:04:52.240 realm of theory for decades. Oh, so it was a 00:04:52.240 --> 00:04:54.660 known problem. Absolutely. A single camera lens, 00:04:55.019 --> 00:04:57.779 no matter how high the resolution, is inherently 00:04:57.779 --> 00:05:00.500 cyclolepic. It has zero native depth perception. 00:05:00.699 --> 00:05:04.000 Right, because a large car far away looks geometrically 00:05:04.000 --> 00:05:07.000 identical to a small car close up. Exactly. But 00:05:07.000 --> 00:05:09.500 if you look closely at the source text, it specifies 00:05:09.500 --> 00:05:11.639 that this high precision measurement isn't done 00:05:11.639 --> 00:05:14.300 with a standard camera. OK, what is it done with? 00:05:14.579 --> 00:05:17.079 It is achieved through advanced stereoscopic 00:05:17.079 --> 00:05:19.800 imaging techniques. Stereoscopic imaging. So 00:05:19.800 --> 00:05:21.740 we aren't talking about one lens. No. We are 00:05:21.740 --> 00:05:24.839 talking about two. We are talking about replicating 00:05:24.839 --> 00:05:28.240 human biology. It is the exact same biological 00:05:28.240 --> 00:05:30.660 mechanism that allows you to catch a baseball 00:05:30.660 --> 00:05:34.000 or navigate a crowded sidewalk. Let me sketch 00:05:34.000 --> 00:05:36.620 this out mentally for the listener because the 00:05:36.620 --> 00:05:39.540 biological parallel makes the math so much easier 00:05:39.540 --> 00:05:42.279 to grasp. Yeah, go for it. So hold your thumb 00:05:42.279 --> 00:05:45.199 out at arm's length. Okay, do it. Close your 00:05:45.199 --> 00:05:47.120 left eye and align your thumb with something 00:05:47.120 --> 00:05:49.800 on the wall behind it. Say a picture frame. Got 00:05:49.800 --> 00:05:52.259 it. Now without moving your hand, open your left 00:05:52.259 --> 00:05:54.420 eye and close your right eye. The thumb appears 00:05:54.420 --> 00:05:56.819 to jump horizontally across the background. Exactly. 00:05:57.399 --> 00:06:00.660 That visual jump is called parallax. Your left 00:06:00.660 --> 00:06:02.560 eye and your right eye are viewing the world 00:06:02.560 --> 00:06:04.860 from two slightly different angles, separated 00:06:04.860 --> 00:06:07.459 by a couple of inches. Right. Your brain takes 00:06:07.459 --> 00:06:11.279 those two flat 2D images, measures the disparity 00:06:11.279 --> 00:06:13.639 between where objects appear in each eye, and 00:06:13.639 --> 00:06:15.860 instantly stitches them together to create a 00:06:15.860 --> 00:06:18.420 3D map of the world. It's pretty amazing. That's 00:06:18.420 --> 00:06:20.439 how you know how far away your sum is from the 00:06:20.439 --> 00:06:23.300 wall. And VIDAR applies that exact phenomenon 00:06:23.300 --> 00:06:26.319 to traffic enforcement. But instead of neurons, 00:06:26.560 --> 00:06:31.480 it uses pure geometry and silicon. So dual visual 00:06:31.480 --> 00:06:34.680 inputs. Yes. The assistive utilizes two camera 00:06:34.680 --> 00:06:37.759 lenses mounted on that gantry or pole, staring 00:06:37.759 --> 00:06:40.139 at the exact same stretch of asphalt. But wait, 00:06:40.339 --> 00:06:42.500 for the computer to calculate depth from that 00:06:42.500 --> 00:06:44.699 parallax, it has to have a baseline, right? I 00:06:44.699 --> 00:06:47.079 mean, the human brain knows how far apart our 00:06:47.079 --> 00:06:49.560 eyes are. Exactly. How does the VIDAR system 00:06:49.560 --> 00:06:52.660 establish that metric? The baseline is actually 00:06:52.660 --> 00:06:54.879 the most critical variable in the entire system. 00:06:55.060 --> 00:06:58.120 OK. The two lenses on a Vidar unit are rigidly 00:06:58.120 --> 00:07:00.560 mounted at an exact mathematically known distance 00:07:00.560 --> 00:07:02.959 from each other. So they don't move? Never. Let's 00:07:02.959 --> 00:07:05.579 say it's exactly 24 inches. The system software 00:07:05.579 --> 00:07:07.860 is hard -coded with that 24 inch baseline. Gotcha. 00:07:08.180 --> 00:07:10.980 When a car drives into the frame, camera A captures 00:07:10.980 --> 00:07:13.980 an image of the license plate. In that flat image, 00:07:14.279 --> 00:07:16.699 the license plate is located at specific pixel 00:07:16.699 --> 00:07:19.759 coordinates. Let's say pixel 1000 on the x -axis. 00:07:20.079 --> 00:07:22.339 And at that exact same millisecond, camera B 00:07:22.339 --> 00:07:25.680 captures its own image. Yes. But because camera 00:07:25.680 --> 00:07:29.120 B is 24 inches to the right, it sees the license 00:07:29.120 --> 00:07:31.120 plate from a slightly different angle. Precisely. 00:07:31.180 --> 00:07:34.300 So in camera B's flat image, the license plate 00:07:34.300 --> 00:07:38.500 isn't at pixel 1000. It's at like... Pixel 1050. 00:07:38.680 --> 00:07:40.860 And that 50 pixel difference is the disparity. 00:07:41.060 --> 00:07:44.240 Ah, the visual jump. Exactly. Because the software 00:07:44.240 --> 00:07:46.560 knows the exact distance between the two lenses, 00:07:46.839 --> 00:07:49.540 the exact focal length of the optics, and the 00:07:49.540 --> 00:07:52.040 exact pixel disparity between the two flat images, 00:07:52.500 --> 00:07:55.439 it can use basic trigonometry to triangulate 00:07:55.439 --> 00:07:58.319 the exact distance to that license plate in three 00:07:58.319 --> 00:08:01.199 -dimensional space. Wow. It has mapped the z 00:08:01.199 --> 00:08:03.920 -axis purely through observation. Okay, so that 00:08:03.920 --> 00:08:06.060 establishes the depth for one single moment in 00:08:06.060 --> 00:08:09.060 time. Right. Speed is distance over time. So 00:08:09.060 --> 00:08:11.879 the Vidar system has to run that complex stereoscopic 00:08:11.879 --> 00:08:14.500 triangulation over and over again. Frame by frame 00:08:14.500 --> 00:08:16.899 by frame. That's a lot of math. It is. This is 00:08:16.899 --> 00:08:18.959 where the video aspect of video detection and 00:08:18.959 --> 00:08:20.939 ranging comes into play, because video is really 00:08:20.939 --> 00:08:23.040 just sequential photography. Sure. Let's assume 00:08:23.040 --> 00:08:25.439 the Vidar cameras are shooting at 60 frames per 00:08:25.439 --> 00:08:28.000 second. OK. If it's shooting 60 frames a second, 00:08:28.060 --> 00:08:30.740 that means there is a gap of about, what, 16 00:08:30.740 --> 00:08:33.100 .6 milliseconds between every single photograph? 00:08:33.559 --> 00:08:36.940 Exactly. So the car enters the frame. The system 00:08:36.940 --> 00:08:39.980 uses stereoscopic disparity to calculate that 00:08:39.980 --> 00:08:42.940 the bumper is exactly 100 feet away. Right. 16 00:08:42.940 --> 00:08:45.580 .6 milliseconds later, the cameras capture the 00:08:45.580 --> 00:08:48.460 next frame. The system recalculates the disparity. 00:08:49.019 --> 00:08:52.740 The bumper is now, say, 98 feet away. So it knows 00:08:52.740 --> 00:08:55.419 the exact distance the car moved through 3D space, 00:08:55.440 --> 00:08:57.759 and it knows the exact fraction of a second it 00:08:57.759 --> 00:09:00.590 took to cover that distance. and distance divided 00:09:00.590 --> 00:09:03.889 by time equals velocity. That is wild. The system 00:09:03.889 --> 00:09:06.210 calculates your speed without ever emitting a 00:09:06.210 --> 00:09:08.909 single pulse of energy. It just watched the light 00:09:08.909 --> 00:09:10.950 naturally bouncing off your car from two different 00:09:10.950 --> 00:09:13.110 angles and let the software do the heavy lifting. 00:09:13.370 --> 00:09:16.009 Which brings up a massive historical question 00:09:16.009 --> 00:09:18.929 for me. The geometry behind stereoscopic triangulation 00:09:18.929 --> 00:09:21.370 isn't new. I mean, humans have understood parallax 00:09:21.370 --> 00:09:23.830 and trigonometry for centuries. Oh, absolutely. 00:09:24.070 --> 00:09:26.610 So if this passive method is so elegant, if it 00:09:26.610 --> 00:09:28.950 saves us from building complex radar emitters 00:09:28.950 --> 00:09:32.570 and expensive spinning LIDAR lasers, why wasn't 00:09:32.570 --> 00:09:34.809 this the standard from the very beginning of 00:09:34.809 --> 00:09:36.909 traffic enforcement? That is a great question. 00:09:37.169 --> 00:09:39.809 And the answer is the limitation was never the 00:09:39.809 --> 00:09:42.570 math. What was it? The limitation was the silicon. 00:09:42.759 --> 00:09:45.379 The source material actually points us directly 00:09:45.379 --> 00:09:48.080 to the moment this barrier was broken. Oh right, 00:09:48.220 --> 00:09:50.899 the patent. Yes. It cites a specific architectural 00:09:50.899 --> 00:09:56.000 origin for this technology. U .S. patent 8184863B2. 00:09:56.279 --> 00:09:59.240 Titled the Video Speed Detection System. Issued 00:09:59.240 --> 00:10:02.139 to an inventor named Ji Gang Wang on May 22, 00:10:02.320 --> 00:10:05.480 2012. 2012? I mean, that is incredibly recent 00:10:05.480 --> 00:10:07.879 when you think about the history of highway infrastructure. 00:10:08.100 --> 00:10:10.379 Very recent. We tend to view traffic cameras 00:10:10.379 --> 00:10:13.259 as these archaic, bureaucratic monoliths that 00:10:13.259 --> 00:10:16.179 have been around forever. But the specific patented 00:10:16.179 --> 00:10:18.960 capability to enforce speed limits using passive 00:10:18.960 --> 00:10:21.899 stereoscopic video is barely over a decade old. 00:10:21.940 --> 00:10:23.600 Yeah, and when you look at what the system actually 00:10:23.600 --> 00:10:26.039 has to do to replace a radar gun, that 2012 date 00:10:26.039 --> 00:10:28.179 makes perfect sense. Because of the processing 00:10:28.179 --> 00:10:31.100 power. Exactly. Think about the computational 00:10:31.100 --> 00:10:33.639 load we just described. Yeah. The system is pulling 00:10:33.639 --> 00:10:36.600 in two high -definition video feeds simultaneously. 00:10:36.720 --> 00:10:39.620 Right. It has to scan millions of pixels in real 00:10:39.620 --> 00:10:42.059 time, identify the vehicle moving at highway 00:10:42.059 --> 00:10:44.960 speeds, isolate a specific feature like a license 00:10:44.960 --> 00:10:47.799 plate or a headlight. and find that exact same 00:10:47.799 --> 00:10:51.600 feature in the second video feed. Yes. Calculate 00:10:51.600 --> 00:10:54.720 the pixel disparity, triangulate the 3D depth, 00:10:55.080 --> 00:10:57.700 and then do it all again. 16 milliseconds later. 00:10:57.840 --> 00:11:00.179 And it has to do all of that math before the 00:11:00.179 --> 00:11:02.320 car physically drives out of the camera's field 00:11:02.320 --> 00:11:05.559 of view. Right. So in the 1990s or early 2000s, 00:11:05.720 --> 00:11:09.059 trying to process dual HD video feeds frame by 00:11:09.059 --> 00:11:12.460 frame for real -time 3D reconstruction would 00:11:12.460 --> 00:11:14.659 have just melted a conventional roadside processor. 00:11:14.720 --> 00:11:16.480 You would have needed a server rack the size 00:11:16.480 --> 00:11:18.340 of a refrigerator sitting next to the highway. 00:11:18.620 --> 00:11:20.580 Pretty much. The hardware simply couldn't keep 00:11:20.580 --> 00:11:23.070 up with the geometry. Radar and LiDAR were the 00:11:23.070 --> 00:11:25.889 standard because computationally speaking, they 00:11:25.889 --> 00:11:27.929 are relatively simple. Because measuring the 00:11:27.929 --> 00:11:30.190 frequency shift of a bounced radio wave takes 00:11:30.190 --> 00:11:32.429 very little processing power. Exactly. You get 00:11:32.429 --> 00:11:35.289 a number instantly. But by 2012, Moore's law 00:11:35.289 --> 00:11:38.629 had done its work. High -definition digital image 00:11:38.629 --> 00:11:41.710 sensors became incredibly cheap and highly accurate. 00:11:41.950 --> 00:11:43.929 Yeah, they were everywhere. And more importantly, 00:11:44.309 --> 00:11:46.789 microprocessors became fast enough to handle 00:11:46.789 --> 00:11:50.450 the sheer volume of data required for real -time 00:11:50.450 --> 00:11:54.129 disparity mapping. Spot on. Jigging Wang's patent 00:11:54.129 --> 00:11:56.909 represents the exact moment when the processing 00:11:56.909 --> 00:11:59.629 power caught up to the math. Wow. It marks the 00:11:59.629 --> 00:12:02.110 transition where software finally overtook hardware 00:12:02.110 --> 00:12:04.850 in the realm of speed detection. And as the text 00:12:04.850 --> 00:12:08.090 outlines that 2012 invention. didn't languish 00:12:08.090 --> 00:12:10.190 in a patent office somewhere. No, not at all. 00:12:10.309 --> 00:12:13.169 The source explicitly lists its primary applications 00:12:13.169 --> 00:12:15.870 as remote sensing and traffic speed enforcement. 00:12:16.070 --> 00:12:18.389 So this passive stereoscopic observation has 00:12:18.389 --> 00:12:20.649 actively reshaped our physical environment. It 00:12:20.649 --> 00:12:22.929 really has. The text mentions these systems are 00:12:22.929 --> 00:12:25.830 deployed in gantries or on roadside poles. That's 00:12:25.830 --> 00:12:27.970 the physical footprint of this paradigm shift. 00:12:28.129 --> 00:12:30.250 It completely changes the architecture of enforcement. 00:12:30.370 --> 00:12:33.490 How so? Well, because you no longer need heavy 00:12:33.490 --> 00:12:36.669 high voltage equipment to emit radar waves or 00:12:36.559 --> 00:12:39.480 or mechanical spinners for LIDAR, the footprint 00:12:39.480 --> 00:12:42.440 of a speed detection system shrinks drastically. 00:12:42.659 --> 00:12:44.700 Right, it doesn't need to be this massive rig. 00:12:44.860 --> 00:12:47.039 Exactly. It becomes just a pair of lenses in 00:12:47.039 --> 00:12:49.899 a small housing, quietly processing the ambient 00:12:49.899 --> 00:12:52.120 light. So the next time you are driving down 00:12:52.120 --> 00:12:54.259 the interstate, and you pass under one of those 00:12:54.259 --> 00:12:56.879 massive metal gantries stretching over the lanes, 00:12:57.259 --> 00:12:59.500 or you catch a glimpse of a camera cluster on 00:12:59.500 --> 00:13:01.519 a roadside pole out of the corner of your eye, 00:13:01.779 --> 00:13:03.639 I want you to look at it through the lens of 00:13:03.639 --> 00:13:05.779 this deep dive. Yeah, it changes your perspective. 00:13:05.769 --> 00:13:08.269 entirely. Don't just think of it as a dumb security 00:13:08.269 --> 00:13:11.049 camera waiting to take a flat flash photograph 00:13:11.049 --> 00:13:13.690 of your license plate. Because it's not. That 00:13:13.690 --> 00:13:17.389 gray box is doing high level real time calculus 00:13:17.389 --> 00:13:20.070 on the physical space around it. You are driving 00:13:20.070 --> 00:13:23.309 through the field of view of an incredibly sophisticated 00:13:23.309 --> 00:13:25.929 stereoscopic set of eyes. Literally watching 00:13:25.929 --> 00:13:29.080 you. It is silently taking in dual video feeds, 00:13:29.440 --> 00:13:31.840 mapping the pixel disparity of your vehicle against 00:13:31.840 --> 00:13:34.519 the background, reconstructing the three -dimensional 00:13:34.519 --> 00:13:37.220 reality of the highway, and calculating your 00:13:37.220 --> 00:13:40.240 exact trajectory and velocity. It is translating 00:13:40.240 --> 00:13:43.039 the physical world into pure geometry. And doing 00:13:43.039 --> 00:13:45.980 it all passively. It forces a complete reimagining 00:13:45.980 --> 00:13:49.139 of how we measure our environment. So what does 00:13:49.139 --> 00:13:52.309 this all mean? We started with a source detailing 00:13:52.309 --> 00:13:55.409 a rather dry sounding Wikipedia entry about video 00:13:55.409 --> 00:13:58.350 detection and ranging. But what we actually uncovered 00:13:58.350 --> 00:14:00.669 is a fundamental evolution in how technology 00:14:00.669 --> 00:14:03.009 interacts with reality. It really is a massive 00:14:03.009 --> 00:14:05.269 shift. We are witnessing the end of the brute 00:14:05.269 --> 00:14:08.529 force era, an era where we had to actively hurl 00:14:08.529 --> 00:14:10.769 radar waves and laser pulses out into the world 00:14:10.769 --> 00:14:13.230 and wait for them to bounce back just to understand 00:14:13.230 --> 00:14:15.169 how fast something was moving. Yeah, shouting 00:14:15.169 --> 00:14:17.450 into the dark. Right. And we have transitioned 00:14:17.450 --> 00:14:20.639 into an era of elegant passive visual tracking, 00:14:20.980 --> 00:14:22.899 where simply gathering ambient light through 00:14:22.899 --> 00:14:25.679 two lenses is enough to decode the complex physics 00:14:25.679 --> 00:14:28.360 of a 4 ,000 -pound object moving at 80 miles 00:14:28.360 --> 00:14:30.899 an hour. If we connect this to the bigger picture, 00:14:31.720 --> 00:14:34.320 knowing the mechanics behind Vidar completely 00:14:34.320 --> 00:14:36.519 changes how we should interact with our built 00:14:36.519 --> 00:14:38.899 environment. How so? Well, the mundane highway 00:14:38.899 --> 00:14:40.980 commute isn't just a stretch of asphalt anymore. 00:14:41.179 --> 00:14:45.100 It is a live showcase of advanced physics, stereoscopic 00:14:45.100 --> 00:14:47.539 biology, and cutting -edge engineering. That's 00:14:47.539 --> 00:14:49.419 a great way to put it. When you understand the 00:14:49.419 --> 00:14:51.720 baseline math and the frame rate calculations 00:14:51.720 --> 00:14:54.879 happening inside those roadside poles, you realize 00:14:54.879 --> 00:14:57.580 that our infrastructure is becoming increasingly 00:14:57.580 --> 00:15:00.360 perceptive. Perceptive? I like that word. It 00:15:00.360 --> 00:15:02.500 doesn't need to ping us with energy to know we 00:15:02.500 --> 00:15:07.049 are there. It is passively, silently understanding 00:15:07.049 --> 00:15:09.669 the geometry of everything moving through it. 00:15:10.169 --> 00:15:11.870 Increasingly perceptive. I think that is the 00:15:11.870 --> 00:15:14.409 perfect concept to land on. It fits, right. It 00:15:14.409 --> 00:15:16.950 really does. We've spent this entire deep dive 00:15:16.950 --> 00:15:19.610 discussing how stereoscopic technology applies 00:15:19.610 --> 00:15:22.070 to cars on a highway because, well, that is exactly 00:15:22.070 --> 00:15:24.090 where the 2012 patent and the source material 00:15:24.090 --> 00:15:25.929 focused. Right. But I want to leave you with 00:15:25.929 --> 00:15:28.590 a final thought to mull over long after you finish 00:15:28.590 --> 00:15:31.649 listening. Oh, this is a good one. If Stereoscopic 00:15:31.649 --> 00:15:34.590 cameras mounted on a simple roadside pole can 00:15:34.590 --> 00:15:38.110 calculate exact speed, trajectory, and vehicular 00:15:38.110 --> 00:15:41.269 data entirely through passive observation. What 00:15:41.269 --> 00:15:43.409 happens when this passive tracking capability 00:15:43.409 --> 00:15:46.049 shrinks even further? Right, because tech always 00:15:46.049 --> 00:15:48.750 gets smaller. Exactly. What happens when the 00:15:48.750 --> 00:15:51.049 processing power gets even cheaper, the lenses 00:15:51.049 --> 00:15:54.330 get even smaller, and the stereoscopic math scales 00:15:54.330 --> 00:15:58.490 down to track human movement? That's wild to 00:15:58.490 --> 00:16:00.799 think about. How does our world change when the 00:16:00.799 --> 00:16:03.100 environment around us can calculate the speed, 00:16:03.259 --> 00:16:05.440 distance, and trajectory of people moving through 00:16:05.440 --> 00:16:08.639 a public square, a shopping mall, or a simple 00:16:08.639 --> 00:16:11.299 city sidewalk just as effortlessly as it tracks 00:16:11.299 --> 00:16:13.460 a car on the interstate? It completely changes 00:16:13.460 --> 00:16:15.539 privacy, for one. It really does. Just something 00:16:15.539 --> 00:16:17.200 to think about the next time you walk past a 00:16:17.200 --> 00:16:17.419 camera.