Automatic building mapping could help emergency responders
September 27, 2012 5:30 AM   Subscribe

Researchers have developed a backpack that generates building layout map in realtime. (yt) The prototype system automatically maps the wearer’s environment, recognizing movement between floors. It was designed at MIT to be used by emergency responders. Read more.
posted by crunchland (34 comments total) 13 users marked this as a favorite
 
Next up: implementing idbeholda. Give them a few more years for idclip, and emergency rescue will only be constrained by outer map boundaries.
posted by Mayor West at 5:34 AM on September 27, 2012 [10 favorites]


A backpack that generates its own map? Forget emergency responders, Dora the Explorer needs this right now.
posted by Strange Interlude at 5:35 AM on September 27, 2012 [13 favorites]


One would hope that if they are responding to an emergency that the user will be able to move a little faster than this chap.
posted by hornet67 at 5:39 AM on September 27, 2012


Geez, it's like they were playing Zelda and said to themselves, "How can I make this reality?"

I am still hoping for a real-life rupee exploit, though.
posted by inturnaround at 5:40 AM on September 27, 2012 [2 favorites]


This is a great idea. Now they just need to combine it with a realtime position reporting system so that incident commanders can see exactly where everyone is in the building. Right now we have accountability systems that let us see who is (or supposed to be) in the building but not where. If there's a collapse or someone is in distress it can take precious time to locate them; knowing where they are can literally be a life saver.
posted by tommasz at 5:44 AM on September 27, 2012 [1 favorite]


Geez, it's like they were playing Zelda and said to themselves, "How can I make this reality?"

I'm convinced that part of why I love GPS map units so much is that I've played so many video games that I can't comprehend why real life doesn't have an automap.
posted by Pope Guilty at 5:50 AM on September 27, 2012 [4 favorites]


I'm wondering how well those lasers work in smoky rooms.
posted by crunchland at 5:54 AM on September 27, 2012


This is really cool and probably very useful.

However, I feel like we need a new word for "big globs of data that look superficially like things humans make, but aren't really the same". For instance, when I say "map" you get all kinds of mental images that are not captured by this technology. Indication of points of interest (which can vary by context, such as someone looking for wiring problems vs someone looking for the bathroom), scaling/zooming (i.e. different levels of detail and highlighting the right bits when that happens, such as which towns to stop labeling at a given zoom level), etc. These are things a human to do. (Yes, I know they mentioned parsing signage. That both proves my point and is insufficient. It proves my point because humans made those labels and the computer is just getting text input via a camera vs a keyboard. It is insufficient because available signage is another context-dependent thing. Which signs should appear on which maps, for instance? Do the bathroom signs go on the wiring maps?)

And it isn't just maps. Auto-generated docs are an even more familiar version of this problem. It isn't enough to just auto-parse an API definition and print the argument list. It isn't even enough to have the program embed doc strings in each API function and print those out alongside. You need context. You need to understand WHY you would call this function instead of that one. You need to know what a typical chain of calls would look like to perform a certain task.

These aren't problems you can just throw more programming at or use faster computers to solve (unless you've got human-level AI). They are things people want to know about data that are not extractable from the data. And any meta-data you embed is always going to be insufficient because people want all kinds of crazy things.

I don't want to downplay this technology because it really is cool and it really could be useful. I just feel like it's abusing the word "map" to call it that. "Slightly chewed raw location data" might be better, but isn't very short and is not at all general.
posted by DU at 6:04 AM on September 27, 2012 [3 favorites]


"to enable situational awareness...in search and rescue operations"

Or in me trying to find my way around any modern office building. I have almost no sense of direction, being essentially unable to understand the relationship between relative and absolute position. I need this in a jacket now godamnit!
posted by howfar at 6:12 AM on September 27, 2012


MetaFilter: This is really cool and probably very useful. However,
posted by Mooseli at 6:16 AM on September 27, 2012 [12 favorites]


Or in me trying to find my way around any modern office building. I have almost no sense of direction...

I usually have a really good sense of direction, not for NSEW but for "where did I start out from". However, my kids' school is a complete nightmare that I've never been able to figure out despite being there many times. It's some kind of weird fractal design that makes it impossible to tell what "level" you are at (i.e. "is this the hexagon near the main office or a side hexagon?") And they don't put out a human-made map on open house nights, so a computer generated one would be a vast improvement. (And probably better-spelled, if the notes my kids bring home are any indication.)
posted by DU at 6:24 AM on September 27, 2012


I can imagine the deep implications of an auto-map for my job, as well.

"Oh, there goes Donny on his coffee break again"
posted by cacofonie at 6:28 AM on September 27, 2012


I saw a six pack of robots at MakeFaire a few years ago that did basically this same thing. The most interesting part was that they had one up on a table doing recon of the tent they were in, and a full resolution of it's scan on a monitor. It was producing an easily recognizable real time profile of every person in the tent.
posted by Kid Charlemagne at 6:31 AM on September 27, 2012 [1 favorite]


Cavers are salivating over this tech. Mapping, or surveying, a cave is slow, tedious task. Being able to travel normally through a cave and come out with an accurate map would make me very happy.
posted by I'm Doing the Dishes at 6:48 AM on September 27, 2012 [1 favorite]


A backpack that generates its own map? Forget emergency responders, Dora the Explorer needs this right now.

Dear Strange Interlude:

I DO NOT support this plan.

Sincerely,

I'm the map, I'm the map, I'm the map, I'm the map
posted by Benny Andajetz at 6:53 AM on September 27, 2012 [1 favorite]


Best, probably most expensive RPG session ever - get a group in a large building with enough of these and enough smartphones with RFID readers for everyone, mounted on the wrist like a Pip-Boy. The smartphone displays the map as it's discovered, the inventory, messages and images from the GM, etc. The GM RFID-tags props, locations, inventory items, clues, events, NPCs and enemies (played by people in costume, or by practical effects and props, or by pico projected visuals!). When a player picks up an RFID tag it flags the GM, who is hidden away somewhere in the area, tying all systems together, communicating with people via text and images to the phones and audio to headsets, as well as keeping an eye on the players using the images provided by the mapping backpack.
posted by jason_steakums at 7:16 AM on September 27, 2012 [1 favorite]


Mischief managed!
posted by arcticseal at 7:18 AM on September 27, 2012 [1 favorite]


Best, probably most expensive RPG session ever -- Dream Park
posted by crunchland at 7:21 AM on September 27, 2012 [1 favorite]


I'm surprised nobody has said this yet, but –

I guarantee you that there's a room somewhere at Google where they're talking excitedly about how they're going to get a few hundred of these things and start making Indoor Maps a ubiquitous reality. That's assuming they didn't call these guys yesterday and convince them to produce the units (or at least send some plans or something over) already.
posted by koeselitz at 7:45 AM on September 27, 2012 [1 favorite]


I don't imagine the LIDAR would work very well in a building that's on fire, even (and perhaps especially) if they're IR-based. IR isn't that much better at seeing through smoke than visible light (if it was, firefighters would just use NVGs). But maybe they can take the same system and apply it to millimeter-wave radar or backscatter X-rays or some other system that could see through smoke. Although I'm not sure I'd want to wear a backscatter-based system around for very long.

But perhaps the real use for such a system isn't actually in the emergency environment, but before an emergency happens. It's incredibly common for emergency personnel to respond to a building that they don't have accurate plans of -- in fact this is really the rule, not the exception.

But if, say, the Fire Marshall's office had a system like this, and used it as part of their annual inspections, or just brought it around periodically while conducting walkthroughs, then actual responders would have a map of the building interior before entry. That could be a pretty nice asset. Even if it's not something that actually worked in real-time while inside the building, just being able to look over a recently-generated map while planning the entry/attack would be pretty neat.

Also, I think you could eliminate the inertial sensor and replace it with a local GPS-like system. John Deere has some enhanced GPS systems that work by using a base station at a known point, which acts like a little GPS satellite. You can get your distance to the base station with very high accuracy (centimeter scale). The problem normally is fixing the position of these base stations really well, but if you didn't care about absolute accuracy, just relative (for the purposes of constructing a map), you could drop a pair of base stations a set distance apart outside the building, and then know your position inside with pretty good accuracy.
posted by Kadin2048 at 8:01 AM on September 27, 2012 [1 favorite]


Like mounting one to the front and back of a firetruck? Or does it need a bigger distance between the two base stations?
posted by crunchland at 8:08 AM on September 27, 2012


Neat idea, and an ambitious mission.

In addition to the challenge presented by a smoky interior, I would expect the barometer to be in for a rough ride. Regions within a building on fire, or with interior fire doors activated, can deviate significantly from the vented pressure profile the SLAM gizmo would need to assume. Particularly staircases.

Add to the wish list: integrate the output from a gas meter.

Seems to me using this in emergency response (via pre-planning or during an actual scene) could raise some interesting liability questions.
posted by maniabug at 8:25 AM on September 27, 2012


Great, now we can raise a generation of people who not only cannot navigate their city without a smartphone, but who get lost in their own house without a smartphone.
posted by entropicamericana at 8:25 AM on September 27, 2012 [1 favorite]


Also, I think it's quaint how they mention situational awareness for emergency responders but don't touch on the completely obvious military applications

Very last line of the article:
Both the U.S. Air Force and the Office of Naval Research supported the work.
posted by Kabanos at 9:32 AM on September 27, 2012 [1 favorite]


Another possible annoyance in a real-world application might be other moving objects, people in particular, causing distortions on the map.

Still, this is very cool and something I, too, would definitely like to see in consumer technology. Just yesterday, I was wandering around the aisles of a large, unfamiliar supermarket and wished I could just pull out my smartphone, launch the OpenStreetMap viewer and navigate the place instead of moving in circles (and looking for another employee because I just pestered that guy over there).
posted by wachhundfisch at 9:53 AM on September 27, 2012


You find yourself in a room ....
posted by benito.strauss at 9:58 AM on September 27, 2012


There was a time I would have gladly paid a million zorkmids for this.
posted by roger ackroyd at 10:02 AM on September 27, 2012


I like how the maps are 1980s hand drawn dungeon - analog looking, almost artistic.

I saw a six pack of robots at MakeFaire a few years ago that did basically this same thing.

The tech was originally designed for rolling robots, but they decided to make it human wearable as the next step.
posted by stbalbach at 10:02 AM on September 27, 2012


Those quadrotor things are super popular now, so why don't they mount it on those? It would almost be like the floating scanners in Prometheus!
posted by orme at 10:20 AM on September 27, 2012 [1 favorite]


Those quadrotor things are super popular now, so why don't they mount it on those? It would almost be like the floating scanners in Prometheus!

Yeah, but didn't Charlize Theron and Stringer Bell just ignore the 3D map and go off for sex?
posted by colie at 10:30 AM on September 27, 2012


Those quadrotor things are super popular now, so why don't they mount it on those? It would almost be like the floating scanners in Prometheus!

And yet we'll still decide to take our helmets off for some stupid reason, and then get lost anyway, and then TOUCH THE ALIEN THINGS!
posted by TheWhiteSkull at 11:22 AM on September 27, 2012 [5 favorites]


Like mounting one to the front and back of a firetruck? Or does it need a bigger distance between the two base stations?

I'm honestly not sure. I think Deere's system was designed to be deployed at the corners of a cornfield, so that you get a pretty big difference in distance as you moved around the field from the two transmitters (they were basically like LORAN). I would assume that as you move the transmitters closer together, the positional error increases. In a multi-storey building you'd really need 3 base stations if you wanted to get elevation as well.

Unfortunately I can't find any information on the JD system online; it seems they've abandoned it in favor of GPS augmentation systems instead. I believe the LORAN-like system was something that predated the Selective Availability shutdown, when there was more interest in alternative high-precision navigation solutions. Still, the principles at work are pretty straightforward and it would probably be much easier to implement today with commodity hardware than it was in the 90s for them.

It seems that most development of this sort of stuff is being done under the general heading of "indoor positioning system", and it includes exactly the type of LORAN-like beacons that I'm (poorly) describing. Based on the Wiki article, apparently it's easier to use Angle Of Arrival rather than Time Of Arrival for position location over a small area... which makes sense, since the difference in arrival time for radio signals over a small area would be very slight. (Though you could compare phase differences, if your transmitters were synchronized closely enough; that's how some LORAN competitors worked.)
posted by Kadin2048 at 1:10 PM on September 27, 2012


Orme, I don't have a link for you, but there have already been papers about just that.
posted by victory_laser at 1:58 PM on September 27, 2012


....and then TOUCH THE ALIEN THINGS!

But it was full of stars and wanted to be my friend!
posted by Kid Charlemagne at 2:42 PM on September 27, 2012 [1 favorite]


« Older Meteorite deity   |   A Boy's Own Broadmoor Newer »


This thread has been archived and is closed to new comments