WEBVTT - Hiding in Augmented Reality

0:00:03.720 --> 0:00:06.720
<v Speaker 1>A breakthrough. An augmented reality does something that appears to

0:00:06.720 --> 0:00:09.840
<v Speaker 1>be simple but in fact is really tricky and lets

0:00:09.880 --> 0:00:13.200
<v Speaker 1>you hide a virtual object behind a real one. I'm

0:00:13.280 --> 0:00:20.440
<v Speaker 1>Jonathan Strickland, and this is tech stuff. Daily. Augmented reality

0:00:20.480 --> 0:00:23.800
<v Speaker 1>involves using some form of display or output device to

0:00:23.920 --> 0:00:27.080
<v Speaker 1>overlay digital information on top of the real data we

0:00:27.200 --> 0:00:30.760
<v Speaker 1>encounter out in the physical world. The typical representation of

0:00:30.800 --> 0:00:34.040
<v Speaker 1>this is some form of goggles or handheld display that

0:00:34.120 --> 0:00:37.239
<v Speaker 1>overlays computer generated imagery on top of a view of

0:00:37.280 --> 0:00:41.519
<v Speaker 1>our physical surroundings. This is just one type of augmented reality,

0:00:41.600 --> 0:00:44.720
<v Speaker 1>but it's a common one. There are numerous smartphone apps

0:00:44.760 --> 0:00:46.879
<v Speaker 1>that will let you look at the world through your

0:00:46.920 --> 0:00:50.280
<v Speaker 1>smartphone screen. You're looking at a live video feed captured

0:00:50.320 --> 0:00:53.400
<v Speaker 1>by the smartphones camera. The app will overlay some sort

0:00:53.440 --> 0:00:56.360
<v Speaker 1>of digital information on top of your live video view

0:00:56.360 --> 0:00:58.959
<v Speaker 1>of the world. It might be directions, such as an

0:00:59.080 --> 0:01:01.120
<v Speaker 1>arrow pointing to the indicating you need to make a

0:01:01.200 --> 0:01:03.760
<v Speaker 1>left turn to get to your pizza parlor. Or it

0:01:03.840 --> 0:01:06.440
<v Speaker 1>might be a virtual character in a game, perhaps a

0:01:06.520 --> 0:01:09.360
<v Speaker 1>big boss you have to battle before you can cross

0:01:09.400 --> 0:01:11.640
<v Speaker 1>the threshold of your local pizza parlor to get a

0:01:11.640 --> 0:01:16.959
<v Speaker 1>pepperoni slice. It's lunchtime. But one challenge a our developers

0:01:17.000 --> 0:01:20.880
<v Speaker 1>have faced is called occlusion. Occlusion in this sense is

0:01:20.880 --> 0:01:25.440
<v Speaker 1>all about an object blocking another object in the real world.

0:01:25.600 --> 0:01:28.679
<v Speaker 1>If one opaque object passes behind another, of course we

0:01:28.760 --> 0:01:30.880
<v Speaker 1>lose sight of it for a moment. For example, if

0:01:30.880 --> 0:01:33.039
<v Speaker 1>we play hide and seek and you go stand behind

0:01:33.040 --> 0:01:34.760
<v Speaker 1>a tree. When I look around, I'm not going to

0:01:34.880 --> 0:01:37.200
<v Speaker 1>see you, assuming of course you're using the tree to

0:01:37.240 --> 0:01:39.280
<v Speaker 1>block my view of you. And if you walk in

0:01:39.319 --> 0:01:41.520
<v Speaker 1>front of the tree, I'll see you, but I won't

0:01:41.520 --> 0:01:42.880
<v Speaker 1>be able to see the parts of the tree you're

0:01:42.920 --> 0:01:45.960
<v Speaker 1>covering up. This is obvious. We experience this every day,

0:01:46.000 --> 0:01:48.520
<v Speaker 1>so I know I'm not sharing anything new with you.

0:01:48.840 --> 0:01:52.640
<v Speaker 1>But now let's imagine we're using an augmented reality application

0:01:52.800 --> 0:01:56.440
<v Speaker 1>and hardware. We have to teach this system to recognize

0:01:56.440 --> 0:01:59.680
<v Speaker 1>the border of real world objects within view. It also

0:01:59.760 --> 0:02:02.919
<v Speaker 1>must recognize the relative distance of those objects from the camera,

0:02:03.040 --> 0:02:05.720
<v Speaker 1>as well as the virtual distance between the display and

0:02:05.760 --> 0:02:08.840
<v Speaker 1>the virtual object. Only then can such a system allow

0:02:08.919 --> 0:02:11.680
<v Speaker 1>for a virtual object to pass behind or in front

0:02:11.760 --> 0:02:14.320
<v Speaker 1>of a real one and have it reflected in the

0:02:14.360 --> 0:02:17.160
<v Speaker 1>display so it looks like it's really happening in front

0:02:17.200 --> 0:02:20.760
<v Speaker 1>of your eyes. This is a non trivial problem. A

0:02:20.840 --> 0:02:24.079
<v Speaker 1>research team at the University of Arizona's College of Optical

0:02:24.160 --> 0:02:28.360
<v Speaker 1>Sciences have a prototype display that tackles this issue. Their

0:02:28.400 --> 0:02:30.920
<v Speaker 1>display allows for a virtual object to be blocked by

0:02:30.919 --> 0:02:33.640
<v Speaker 1>a real world object. It also allows the virtual object

0:02:33.639 --> 0:02:37.240
<v Speaker 1>to block other objects that are behind it. That behind

0:02:37.520 --> 0:02:40.760
<v Speaker 1>was in air quotes, because obviously the virtual object isn't

0:02:40.840 --> 0:02:44.600
<v Speaker 1>really in the physical space at all. According to Professor

0:02:44.680 --> 0:02:47.959
<v Speaker 1>Hong Hua, the display takes the view of the real

0:02:47.960 --> 0:02:52.519
<v Speaker 1>world and with lenses, images it upon a spatial light modulator.

0:02:52.919 --> 0:02:57.359
<v Speaker 1>These are devices that can alter or modulate polarization, amplitude,

0:02:57.400 --> 0:03:00.560
<v Speaker 1>and phase of light beams. They often will modify the

0:03:00.600 --> 0:03:03.400
<v Speaker 1>intensity of a lightbeam, though more advanced ones can do

0:03:03.440 --> 0:03:06.399
<v Speaker 1>all sorts of funky things with light. The team used

0:03:06.400 --> 0:03:09.639
<v Speaker 1>a spatial light modulator to create a mask that blocks

0:03:09.639 --> 0:03:12.160
<v Speaker 1>out the bit of the real world that the virtual

0:03:12.200 --> 0:03:16.160
<v Speaker 1>object will inhabit. This modulated image is what then reaches

0:03:16.240 --> 0:03:20.480
<v Speaker 1>the display that you can see. One thing that's necessary

0:03:20.480 --> 0:03:23.400
<v Speaker 1>to make all this work is a depth sensor. Without

0:03:23.400 --> 0:03:26.359
<v Speaker 1>a depth sensor, the system cannot know which objects are

0:03:26.400 --> 0:03:29.560
<v Speaker 1>closer to the lens and which are further away. There

0:03:29.560 --> 0:03:31.640
<v Speaker 1>needs to be a way to measure that distance and

0:03:31.639 --> 0:03:34.880
<v Speaker 1>then calculate where a virtual object should fall based upon

0:03:34.920 --> 0:03:38.520
<v Speaker 1>that objects virtual distance from the viewer. This is a

0:03:38.560 --> 0:03:41.520
<v Speaker 1>big jump from just overlaying a digital image on top

0:03:41.560 --> 0:03:44.520
<v Speaker 1>of a video view of the world. This is incorporating

0:03:44.560 --> 0:03:48.160
<v Speaker 1>computer generated imagery into a real environment as if it

0:03:48.280 --> 0:03:51.240
<v Speaker 1>too were just as real. Law says that it will

0:03:51.240 --> 0:03:54.160
<v Speaker 1>likely be sometime before we see this capability built into

0:03:54.160 --> 0:03:58.480
<v Speaker 1>consumer headsets. She says her team's solution wasn't exactly sleek

0:03:58.520 --> 0:04:00.800
<v Speaker 1>and sexy. They were just trying to get a system

0:04:00.840 --> 0:04:04.160
<v Speaker 1>that worked. Refining that system into something more consumer friendly

0:04:04.240 --> 0:04:06.320
<v Speaker 1>will take some time to get it down to a

0:04:06.360 --> 0:04:09.080
<v Speaker 1>format that might fit in, say a pair of special

0:04:09.160 --> 0:04:13.280
<v Speaker 1>glasses will take even more time, but the foundation is there.

0:04:13.600 --> 0:04:17.440
<v Speaker 1>To learn more about augmented reality, mixed reality, and all

0:04:17.480 --> 0:04:20.279
<v Speaker 1>sorts of other types of reality. Subscribe to the Tech

0:04:20.320 --> 0:04:23.599
<v Speaker 1>Stuff podcast every Wednesday and Friday. I look at big

0:04:23.640 --> 0:04:26.279
<v Speaker 1>topics and tech and really explore how they work and

0:04:26.320 --> 0:04:28.880
<v Speaker 1>why they're important. I'll see you again soon.