I have a fantasy that someone reverse-engineers the tape data format and is able to render new maps to it. For the 2 etak systems still operational out there...
I have several of the gyro and inclinometer units, and the magnetometer compass, somewhere. The gyro was a motor spinning a flexible metal plate. As the vehicle moved, the plate would flex, and sensing the plate's position gave a rough turn rate. The inclinometer was a little sealed cup with four capacitive sensing plates. I was looking into using this for a robotics application, but it was too big for an R/C car sized vehicle.
I'm surprised they had a gyro (the article also only mentions a compass, which makes sense to remove accumulated errors in heading). I would have expected the wheel sensors to provide data of similar quality as a gyro back then, without the cost.
The style is an artifact/limitation of using a vector display.
Compare to today’s ubiquity of a raster display. Why did they choose to use a vector display? Maybe to decrease cost and avoid placing framebuffer memory? Maybe rendering maps directly to a vector display could be faster by skipping a rasterization process? Any other reason?
Maps are intrinsically vector data, and a raster graphics display back then would have been low-res, 320x240 at most, making the map (and text!) really difficult to read. And then you’d need the rasterizer itself, using precious CPU cycles and memory bandwidth to turn perfect mathematical line segments into crude pixelated approximations. And yes, the memory needed for the framebuffer was also likely an issue. The question is more, why would they ever have used a raster monitor? None of the advantages of raster were applicable, and the disadvantages were all relevant in their use case. The 100% obvious choice was vector.
I believe a vector CRT could be color just like a raster CRT can, using three phosphors and three electron beams (sure, technically that would’ve made the monitor a vector-raster hybrid). That would’ve raised the cost even higher, which I assume was the main reason the system was monochrome. Sure, you couldn’t easily render filled geometry with a vector display, but it wouldn’t have been anywhere near feasible with a raster monitor either given the puny hardware.
Perhaps one of the most widespread device that used a vector display was the original Asteroids arcade game.
There’s a functioning machine at the Alamo Drafthouse in San Francisco (at least there was last time I went there), I can’t help but stare at it (and give a mini display history lesson to friends) every time I go. Those lines are just so crisp and bright and beautiful.
Yeah, it's always so cool looking. The device is using a CRT vector display, so instead of the CRT drawing each pixel row line by line, each shape on the screen is drawn one by one as small line segments.
Curves are also possible, but you'd have to formulate the vector shape for it, which is harder than for straight lines.
It also looks even cooler in person, as the refresh rate is also really good due to the CRTs, if there's an old arcade close with Asteroids or similar early vector games I'd really recommend going to see it.
There were CRT based in-car record keeping systems used in police cars for a long time before laptops replaced them which had a very similar form factor.
It is amazing that they got it to work as well as it did given how it was 15-20 years ahead of its time. Sure the unit ended up costing as much as the car it was mounted in, but given the limitations of the technology of the time that is simply amazing. To get an idea of how ahead of its time this is, it wouldn't be until a year after the release that High Sierra formatting for CD-ROMs would be proposed. A CD Drive would have added even more expense, but it should would have beat out swapping around dozens of cassettes.
I wonder how much memory it had. The contemporary PC-XT using the same chip started out with 128kb but could be expanded to 640kb. One can imagine it had to page data in and out of that slow cassette interface quite regularly as you're driving around.
a word of Micronesian origin for a distinctive cognitive and mnemonic approach to oceanic navigation and orientation involving a notional reference point or "island", called etak, and triangulation based on it.
... the use of a relative frame, in which the boat is considered to be at rest, while the etak moves.
The name is perfect. "Etak" refers to a system of navigation used by Micronesian and Polynesian seafarers to navigate from island to island in the vast Pacific Ocean. Much like this device, it operates by "augmented dead reckoning" (as it says in TFA).
The etak system of navigation involves navigating by stars and ocean swells to get the heading, but a key issue when navigating by dead reckoning over long distances is that if you're a fraction of a degree off you may miss your destination and never know it, so it's also vitally important to know how far you've gone. This is hard when there are no landmarks. The navigators estimated their distance by using intermediary islands off to the side, which they viewed as coming towards them (from their point of view, the navigator stays motionless on the open ocean while the world moves towards them) and past them. These reference islands were called etaks.
However, confusingly, the etaks were generally not visible, being beyond the horizon, and sometimes did not even exist. The navigators would have named etaks that they pictured being just over the horizon, whether they were there or not, and would track their procession past their boat. When the set number of etaks had passed, they would know they were in the vicinity of the destination island. If they were not at the right time of day for birds to be out, they would then hang out in the area waiting to spot the birds leaving or returning at dawn or dusk.
So the system involves dead reckoning plus a system of turning the navigators' own well-developed intuition of how far they had travelled into a formalized system of generally-invisible islands that they used as a mental model to externalize this intuition.
(My knowledge of this is from Cognition in the wild, Hutchins, E., 1995.)
Edit: D'oh, I should have finished TFA. This is described at the end, although more roughly.
This sounds very romantic, but they were mostly at drift. Most very using rafts without any form of propulsion. "Navigation" across wast distances was one way road with no return ticket. They had to do it for overpopulation, not for some explorative spirit.
I don't have references easily to hand but I don't think that's true. Polynesian navigators were able to make repeat journeys to the same destination reliably enough to return home and build maps and train successors in their craft.
There is plenty of evidence that Pacific Islanders were able to navigate from island to island, for example in the book I cited above. Here's more [1]. What's your evidence for them not being able to?
Honestly, this sounds like some crap spread by people who can't bring themselves to believe that non-Western "primitive people" could have highly developed skills.
The Polynesians used double-hulled canoes not rafts. King Kamehameha conquered the Hawaiian island using giant canoes with accounts from Westerners.
The system of navigation was recorded orally. People in living memory retained the knowledge. Anthropologists got it from them, built replica canoes, and then sailed around discovering that it worked. Captain Cook used Polynesian navigator who was well traveled.
Maybe the rafts are coming from Easter Island losing the technology and using rafts instead.
It only happened in the late middle ages, after possibly millenia of experience navigating easier waters. They simply didn't teach their methods to anyone
Another book specifically about the Polynesian navigators is The Last Navigator. Having no experience navigating at sea, parts of it were a bit over my head, but it was a great read. It also goes into the culture of the people who are the subject of the book.
I've read somewhere that part of the method they intuited their way was to read the waves. As faraway land masses can affect the shape of waves, supposedly these navigators could "see" beyond the horizon due to how the swell was behaving.
I can sort of see that in a mind's eye, with rings of waves spreading as they bounce off obstacles in water. But that's bird's view of a miniature -- and seeing that from the surface would be a very different story.
I’m reminded of something that I heard someone say once in an interview—intuition is just memory in disguise. Certainly, in some areas of my life, my strategy is to learn things and forget them to be able to use them. It doesn’t work in all arenas (mathematics really demands explicit remembering much of the time, but I find programming and writing both work well with forgotten remembering).
As a young child I went mackerel fishing on a small boat in Shetland with my father and a local he knew.
The man steered the boat with one had in the water, until he announced we were near a shoal of mackerel - he said he could feel the grease of their bodies in the sea water.
We threw some lines in and sure enough we caught plenty of fish for dinner.
I used to fish for salmon quite a bit. Herring schools release enough oil into the water to visibly affect the wave action at the surface. You can also smell it.
I heard that polynesians, when asked (in the context of setting out on deadly journeys in wooden boats) "how did you know there would be land there" said "we read the waves".
They could see from the patterns in the waves, thousands of miles away, where the land masses were.
>insert joke about training neural networks
(If anyone knows the original quotes, please post below.)
That sounds fascinating, but it's not really clear to me how imagining islands beyond the horizon can help with dead reckoning. Maybe there are changes in observable phenomena, such as ocean currents, that are associated with these unseen islands? It does sound like a very complex system based on the beginning of this article; I'm wondering if anyone here has read the books mentioned in it: https://www.jstor.org/stable/20705519
I've found a bit more information about the topic in a publicly available research article [1]:
"A depth of only 25 fathoms is quite enough to give some surface indications: coloration, wave phenomena, perhaps fauna. Is this the explanation of the ghost island? Some lost traveller, perhaps en route from Yap to Guam, seeing and remembering these phenomena, later reifying them as an inhabited land? Or is it possible that a real island once existed here, as the Carolinians say? [...] Any Carolinian navigator worthy of the name can give a whole set of radiating courses under all the navigation stars from every island of the Carolines, not just from Kaafiror. [... N]avigators do learn them, together with the courses from real islands, and they make no distinction among them. It is perhaps not altogether in the realm of fantasy to speculate that the curriculum of the schools of navigation was established in a time when Kaafifor was more than a discolored patch of water."
Yeah, this surprised me as well. I get how reading ocean swells, sea life, birds, ocean color (indicating depth and/or plant life below) could give a general sense of position. And I can see how stars, prevailing currents and maybe even estimates based on the relative movement of clouds (adjusted for ambient wind direction and weather conditions) could give a general sense of distance traveled. But on a completely cloudless day (or fully overcast night) gauging distance traveled seems like it could be catastrophically imprecise often enough to make for short navigation careers.
I'm not sure how the concept of tracking virtual islands over the horizon really helps. The only thing I can think of is maybe the idea of it encourages the navigator to stay focused on estimating the passage of proxy points on the far horizon based on whatever composite of wind, current and swell signs they are intuiting from. While still quite variable, I assume gauging distance estimates on the far horizon is better than the alternative of trying to estimate distance traveled from the immediate surroundings of the craft (which are only useful for estimating velocity).
> I'm not sure how the concept of tracking virtual islands over the horizon really helps. The only thing I can think of is maybe the idea of it encourages the navigator to stay focused on estimating the passage of proxy points on the far horizon
You can practice navagating against physical islands over the horizon, and when you're good at that, you've mostly gotten good at dead reckoning against a real reference point; of course, with corrections from the islands influence. Having a community shared archipelego of virtual islands lets you focus your dead reckoning skills on a point while offering a vocabulary of distance and reducing travel times between waypoints.
Go 1000 miles in this direction seems a lot harder for me to follow over many days than go X miles to A, then Y miles to B, then Z miles to C. Even if A and B aren't real. If I treat them as very small islands that will be over the horizon, no big deal that they don't influence the environment, they're small; but I can't really use them to course correct, my reckoning needs to be good.
The explanation could be a combination of experience, details, survivor bias and true scottman.
That is with experience you can select something on the ocean far away that you can track (kelp, etc), with experience and focus to accurately track it and take into account its own movement. Then the method is obviously only promoted by those who successfully survived using it, as the potential nay-sayers who used it and failed are no longer there to give a counter-point. Finally, those who did not use it successfully are probably characterized as "not good navigators", in circular logic.
We had a ubiquitous continental data network back then. You can see it in operation in the old historical film, "You've Got Mail". The floppy installation disks were so plentiful they probably form a sedimentary stratum future geologists will use to identify the era.
That was 1995. In 1991, we had the ubiquitous intercontinental phone network, but overseas connections were >$1/minute and data rate was only 9600 baud.
There was a satellite ISP for a time. The satellite box had a modem in it and your outbound traffic went out over the modem and the inbound traffic came down to the dish.
Still, email and usenet sort of worked. I have sent my first international email in 1993, from a school lab in my home town in Siberia. In larger cities, I bet there were ways to do it even earlier.
Your last line immediately made me think. probably in near future, techies of the future would say something similar about our achievements, maybe gpt, or 4 qbit quantum computer.
Me too. Atari, toward the end, had even mastered color vector displays that look excellent!
Look for clips of "Star Wars" in action.
The 70's was anni interesting time. Atari was employing dynamic vector displays capable of real time motion.
Tektronix was using vectors with their storage CRT tech. Basically the vectors got painted onto the tube phosphors, thus displaying the image without the need to refresh.
4k resolution (vector coordinate space) ended up being a thing!
Highly inspiring! Will share with my team (and I don't do that often!)
It seems that Etak was to navigation systems what Jodorowsky's Dune was to 1980s sci-fi: a trail blazing endeavour that was wild and wildly innovative, did not fulfill its intended mission but rather set up an entire field for subsequent success.
Also: the design must have included several masterpieces when considering the state of tech in the 1980s: even seeking to the right point on the map cassette is an untrivially hard problem.
I concur with you that Etak delivered a functional product, albeit not a commercially successful one, which is farther along than Jodorowsky's Dune project, which stalled mid-way in development.
What I wanted to point out is that both projects produced massive amounts of reusable knowledge, and that knowledge set up the stage for a whole field of influential and exciting derivative works.
> I concur with you that Etak delivered a functional product, albeit not a commercially successful one, which is farther along than Jodorowsky's Dune project, which stalled mid-way in development.
You seem to be responding to another comment? Or are putting words in the commenters’ mouth?
I was responding to the "Jodorowsky’s attempt was ... never released" critique by trying to expand on both the similarities and dissimilarities between Etak and Jodorowsky's Dune.
My comment wasn't really about Etak at all, I just thought you might have mixed up Jodorowsky's Dune with Lynch's! The Lynch one seemed like a better fit, although I think "set up an entire field for subsequent success" is a bit of a stretch for both of them.
The documentary “Jodorowsky’s Dune” presents a fairly strong argument that the “Dune bible” that was assembled for that movie and shopped around to production studios strongly influenced many of the great sci-fi films (or at least iconic scenes in them) in the years to follow.
Specific examples from the documentary were Star Wars and Alien. The latter of which included Dan O’Bannon (screenwriter) and H.R. Giger (effects+concept artist), reprising their roles as staff on Jodorowsky’s Dune.
Highly recommend the documentary by the way. I actually just saw it a couple weeks ago, so the comparison made a lot of sense to me.
Aha, thanks, that’s interesting! I knew some of that but wasn’t aware it had influenced Star Wars. That does make some sense.
Edit to add: I’d still unquestionably put Star Wars as the one that “set up an entire field for success”. Not just because it was the one that shipped and was a massive hit, but because it synthesised many elements besides Dune. I’ll watch that documentary with an open mind though.
Stan Honey other claim to fame (other than being literally the best yacht navigator, probably ever) was founding Sportsvision, the company that created the yellow 1st down line you see when you watch football on tv.
Back in the day the first “computer graphics” class we had at uni was on a tektronix 4010. You would build 3D models and rotate them and display the movement on the monitor. This was when all screens were green text only in a time share system.
> When I worked on the Apple Maps team, 12 of my colleagues were Etak alumni.
What a legacy! It's gratifying to hear of these long-dead companies/products with incredible engineers who are still out there slinging code with the best of them.
Stan Honey, heavily mentioned in the article, most famous for his contributions to sailing, also built the yellow first down line graphic were used to seeing in American football.
>Every once in a while, a revolutionary product comes along that changes everything. Or so said Steve Jobs when he announced iPhone in 2007.
Interestingly, one of Etak co-founders was Nolan Bushnell, and he was Atari co-founder that hired Steve Jobs (or specifically Allan Alcorn) in his only full-time job prior to Apple, pardon the pun.
A very long time ago (~1996-1997) I worked for a company in Sunnyvale called Vicinity which made mapblast, a website now so lost to time that google will assume you are typoing mapquest if you search for it. The company was eventually sold (well after I had left) to Microsoft and became part of MSN I believe.
The tech behind mapblast also powered the first version of Yahoo Maps, which was a pretty big deal at the time (this was before Google Maps eventually came in and overshadowed everyone else in that market).
I was in my early 20s and working for one of my first tech startups at the time but Vicinity was primarily made up of graybeards who had previously worked for Etak (who was the primary map data provider for the online mapping system we made, so they had a lot of experience with it) and many of those people were also ex-Atari prior to Etak.
> To solve this problem Etak invented ‘augmented dead reckoning’. This used a process to match the position given by the navigation sensors to a topologically correct electronic map. Whenever the vehicle turned you made the assumption that you’re driving on a road. At that point the location could be ‘snapped’ back to the road and the error from the sensors could be reset. This technique was later adopted by all navigation apps and is still in use today.
No way did they invent this. Not even close!
This is called map matching. It predates Etak by at least 20 years, if not more.
This paper was published a decade before which does exactly this: Lezniak TW, Lewis RW, Mcmillen RA. A dead reckoning/map correlation system for automatic vehicle tracking. IEEE Transactions on Vehicular Technology. 1977 Feb;26(1):47-60.
This was my sense as well. The device looks and is described like something I would not be surprised to find on a 1970s warship or spy plane to aid navigation. Not with street maps, specifically, of course, but something similar!
Still impressive to get it into a consumer-sized (and almost consumer-priced) box.
I mean, if that paper wasn't declassified until 2017, what are they chances they knew about the prior art exactly?
I'm not disputing your assertion, but perhaps I'm a little more charitable in thinking they could have independently invented the same thing and believed they were the first since the one you mentioned was apparently classified.
As an aside, wanting me to pay $33 to read a PDF of a paper from 46 years ago is... unfortunate. (I have a list of other words I'd rather use, but I'm being civil)
One thing I notice about the 80s is that people were much more willing to pay top dollar for first, very limited versions of products. $4000 in todays money is almost the same as Apple Vision Pro, for a product that has very limited usability.
May be it was easier to market only for rich people who wish to show off then? Since the fall of Vertu no tech companies seem to address specifically this segment. Or may be people just were more optimistic about tech?
Is that true, though? Etak had to license their tech and court a buyer shortly after going to market. Meanwhile, Apple has sold over 200k Vision Pro headsets.
> One thing I notice about the 80s is that people were much more willing to pay top dollar for first, very limited versions of products.
I'd rather think that there is more money frivolously spent today; in the S.F. Bay Area, much more.
> May be it was easier to market only for rich people who wish to show off then?
I'm quite perplexed about how your perception can be so very different from mine. How many people own a Tesla in your neighborhood?
This device however, I would have thought, would have been marketed chiefly to professionals. Traveling salesmen, doctors (who then still made house calls), service technicians etc. .
> The second key invention was a ‘heading up’, moving map display. This meant that the vehicle remained at the center of the screen and the map moved and turned under the vehicle. What you saw ahead of you in the windshield was what was displayed on the screen. This proved highly intuitive.
Later on they talk about "heads up" map digitizing, did this mean the map rotated as the operator digitized the street? Seems quite unwieldy (and how did the poor PC rotate raster graphics?)
I don't think so... I think heads up for the digitizing indicates that the digital map was overlayed on top of the (scanned) source image? As opposed to digitizing from a paper map where you have the map on a surface in front of you, and a digital map hopefully on a screen in front of you (but they did say some were digitizing blind before this?) and you're trying to get the digitized version right by looking between the two.
For the in car map, a vector CRT and vector data makes rotation reasonable. Much less hard than rotating rasterized scanned images on a PC with no rotation acceleration.
I’m guessing the 'blind' method looked something like using one of the early graphics tablets to trace routes or tap control points on a paper map according to a sequence displayed on a text terminal, with no graphical feedback to confirm the vector data during input.
in 1988 i was doing an internship at a company designing ship propulsion systems. they had a CAD computer with with a huge screen and tablet. not sure how old that device was or how expensive but i guess in '85 the technology was not far away.
digitizing blind at that point would mostly be used because it was cheaper than getting a graphics capable computer.
using a projector instead of a screen would be enough to devise a system where the digital image is shown on top of a printed map. so when they came up with that idea they probably already had most of the pieces they need to make it work.
Yes, the map rotated in real time with the vehicle centered. I was a software developer writing software using Etak in '88-'89 time frame. That only worked when focused on a single vehicle, as I remember writing the code to do the same for groups of vehicles and Etak wanted to purchase that code from my employer. they probably got it, as my lead developer I worked for ended up working at Etak after I left.
Interesting side-fact: we used time-of-flight with beepers placed in the cars and ordinary trigonometry to increase accuracy. Worked like a charm.
Heads up map digitizing seem to refer to the method where computer operator digitized an aerial image map by manually tagging its features displayed on the screen, in contrast to heads down digitizing where they used a special tablet.
Anyway I really need such articles from time to time not to lose my faith in humanity.
Rotating raster graphics is easy and the technique was well known at the time, see for example https://www.ocf.berkeley.edu/~fricke/projects/israel/paeth/r... . A PC was easily capable of it. Maybe not in real time as we're used to today, but quick enough that it wouldn't be an impediment.
Heads up digitizing means digitizing the road vectors on top of an image. In Etak's case we used 5 bits to display the image and 3 bits to display the vectors on top. Prior to this invention people used a method of tablet digitizing (similar to https://www.wacom.com/en-us/products/pen-tablets) which was much slower and more error prone (and thus less accurate).
"The prototype driver navigation system was unveiled in Hildesheim on June 21, 1983, and it proved groundbreaking: EVA was the first ever experimental autonomous navigation system."
Very interesting, i've just yesterday wrote about the successor of EVA, the TravelPilot IDS [1] which was commercially available. But I didn't knew that there was another system 5 years ahead.
Both (Etak and TravelPilot IDS) seems to use kind of a vector display. Does someone know if this is for better resolution or better contrast, or both or if there is another reason?
There is no conventional resolution to vector displays (though monitors have limitations).
Reminds me at one point in time I owned a Vectrex,
which had much cleaner lines than any other console,
even much better than my later, much more expensive Amiga
or (early) PCs.
I worked at the company that made the Vectrex, Jay Smith was a wonderful guy. Super brilliant too: the world of PSX Bowling games was his creation, he wrote the first bowling physics sim using Excel, and we hired some guy with a PhD in Statistics who wrote Jay's algorithm in assembly for the PSX, and the 3D bowling genre was launched.
I wonder what was the state of navigation systems for planes at this time, because among the car GPS brands, one finds for instance Garmin, which is an avionics company.
The article claims that the "match roads by turns" technique "was later adopted by all navigation apps". Does anyone know if this is true? My impression was that they rely on GPS position only for positioning, even though modern phone hardware should give really nice gyroscope/accelerometer data.
Virtually all car navigation software, or phone GPS apps in driving mode, will "snap" to a nearby road if the GPS indicates that you are traveling parallel to one. This compensates for minor GPS reception errors.
This can occasionally, in rare situations, be a problem, if you have frontage roads very close to a highway, they can sometimes get confused about which road you are actually on.
In car navigation software it's not all that rare. Chances are the maps are out of date, perhaps by quite a few years (because manufacturers ask for absurd prices on map updates) and you're traveling on a road which doesn't yet exist on the map ...
I love when I've been traveling at highway speed down a highway for over an hour, and suddenly my GPS starts giving me directions back to the highway from some nearby parallel road when I haven't so much as passed an exit.
It's incessant for me; a significant part of one of my frequent routes is on the service-drive that parallels a major freeway, after taking an exit, but prior to diverging. Depending on traffic, though, it may also be advantageous to stay on the freeway, so both are valid parts of the route.
If Waze instructs me to take the exit, then it assumes I'm on the service drive, even if GPS says I'm still on the freeway. And vice-versa, more problematically -- if I impulsively take the exit, it assumes I'm still on the freeway even if GPS clearly shows I'm on the service drive.
(I can confirm this by running a spare laptop with a USB GPS as a logger, while my iPhone runs Waze. Overlay the GPX on a map later and it's super obvious whether I took the exit or not, but either the Apple location provider or Waze staunchly ignores reality in favor of obsessive road snapping.)
Where this gets stupid is, if there's a traffic jam on the freeway and I dip onto the exit to avoid it, now Waze sees me flowing freely down the service drive, assumes that it's the freeway that's flowing freely, and disbelieves other users who report traffic there. Even as the service drive curves and diverges and I follow the curve, it doesn't retroactively say "Oh jeez, he must be on the service drive after all, adjust the previous data to apply to the service drive and not to the freeway!". So the bad data continues to corrupt the traffic picture and encourage other users to get stuck in traffic they can't report.
I interpreted that to mean that any error in GPS coordinates will be snapped to the closest road that matches the direction vector. (At the same time, I doubt my understanding since I've seen plenty of navigation systems show the vehicle not on the road when traveling on less well covered GPS areas).
GPS and other signals aren't continuous, they all use dead reckoning to fill in the blanks in between. This was even more of a necessity with early smartphones and navigation systems that only had GPS; nowadays they can use a combination of GPS, GSS, Gallileo, GPS and wifi networks. The latter was a secondary goal of the Google Street View project, matching GPS / location with wifi signals.
When nearing a turn you're supposed to take, try stopping completely and rotating your phone as if you're in the turn. Google Maps will continue into the turn before correcting back to the place you stopped. It doesn't do this if you don't rotate the phone.
> Compact tape drive. which uses 1/4-in. magnetic tape cassettes operating at 80 ips, each containing every street and specific address, for an area about twice that of an ordinary paper street map, as well as overviews of major state and regional roads, and national interstates (installed under the vehicle dashboard or in the glove compartment).
So it seems that, if it's not 80 inches per second, then the confusion dates at least back to 1985!
"The actual production of Musicassettes was done on machines running 32 times faster than normal playback. Cassette tape would be reeled over four heads recording what would be both sides at once at 60 IPS. The master tape that was source of the original music had been recorded at 7.5 IPS and this would also run 32 times faster, clocking up a playback speed of 240 IPS for duplication purposes."
"This super-fast tape transport also required the circuitry to follow suit. So instead of the bias frequency being around 80kHz, it was now 2.4MHz; the amplifiers also needed to work over a frequency range of 200kHz to 500kHz."
Commodore 64 with best tape Turbo is able to store slightly above 1MB per cassette. Japan already had floppy drives capable of storing over 1MB in 1983, but it looks like Etak needed more, with this 40 times faster tape drive delivering:
"local map data base stored on a 3.5-MByte tape cassette."
I would love to learn more details about this drive. Modulation used? Number of tracks? Format? Magnetic flux dump of one of the cassettes would be a lovely puzzle to decode.
I had a friend who had one of these in the mid-90's. It was pretty cool (at the time)
I remember installation wasn't trivial. It needed a lot of futzing with the car. I remember the wheel rotation sensors, and they are briefly mentioned in the article.
Smazing that nowadays all this stuff is solid state and in your pocket.
None of this stuff is solid state. What's solid state is completely different stuff, namely GPS. There's also a gyroscope, accelerometer, and compass, but I don't know if those can be considered solid state.
When I started working at TeleAtlas Germany (former Robert Bosch Data; forever a part of TomTom now) in 2005 we still had production processes on the MapEngine technology coming from Etak. We had in-house python bindings that allowed for very productive development. It is fun to see this mentioned here today.
I'm interested in this kind of thing. How things work and so on, the mechanisms and algorithms behind it. The article is long and I haven't read the whole thing yet, but I upvoted it to read it later. That way it's easy to access.
211 comments
[ 4.0 ms ] story [ 250 ms ] threadhttps://www.computerhistory.org/collections/catalog/10276659...
Etak Navigator Tour and Demo [video] - https://news.ycombinator.com/item?id=32202425 - July 2022 (1 comment)
Who Needs GPS? The Story of Etak's 1985 Car Navigation System - https://news.ycombinator.com/item?id=13744825 - Feb 2017 (83 comments)
https://en.wikipedia.org/wiki/Vector_monitor
Compare to today’s ubiquity of a raster display. Why did they choose to use a vector display? Maybe to decrease cost and avoid placing framebuffer memory? Maybe rendering maps directly to a vector display could be faster by skipping a rasterization process? Any other reason?
> None of the advantages of raster were applicable
Colour might have been nice though.
Perhaps one of the most widespread device that used a vector display was the original Asteroids arcade game.
There’s a functioning machine at the Alamo Drafthouse in San Francisco (at least there was last time I went there), I can’t help but stare at it (and give a mini display history lesson to friends) every time I go. Those lines are just so crisp and bright and beautiful.
It also looks even cooler in person, as the refresh rate is also really good due to the CRTs, if there's an old arcade close with Asteroids or similar early vector games I'd really recommend going to see it.
I wonder how much memory it had. The contemporary PC-XT using the same chip started out with 128kb but could be expanded to 640kb. One can imagine it had to page data in and out of that slow cassette interface quite regularly as you're driving around.
Interesting that it cost about as much as the Apple Vision Pro. Probably had way more utility for its buyers.
[0] https://www.michigan.gov/-/media/Project/Websites/sos/01hile...
The etak system of navigation involves navigating by stars and ocean swells to get the heading, but a key issue when navigating by dead reckoning over long distances is that if you're a fraction of a degree off you may miss your destination and never know it, so it's also vitally important to know how far you've gone. This is hard when there are no landmarks. The navigators estimated their distance by using intermediary islands off to the side, which they viewed as coming towards them (from their point of view, the navigator stays motionless on the open ocean while the world moves towards them) and past them. These reference islands were called etaks.
However, confusingly, the etaks were generally not visible, being beyond the horizon, and sometimes did not even exist. The navigators would have named etaks that they pictured being just over the horizon, whether they were there or not, and would track their procession past their boat. When the set number of etaks had passed, they would know they were in the vicinity of the destination island. If they were not at the right time of day for birds to be out, they would then hang out in the area waiting to spot the birds leaving or returning at dawn or dusk.
So the system involves dead reckoning plus a system of turning the navigators' own well-developed intuition of how far they had travelled into a formalized system of generally-invisible islands that they used as a mental model to externalize this intuition.
(My knowledge of this is from Cognition in the wild, Hutchins, E., 1995.)
Edit: D'oh, I should have finished TFA. This is described at the end, although more roughly.
Felt the article was heavy on ad copy and graphics.
see also https://en.wikipedia.org/wiki/Hōkūleʻa
Navigation methods aside, those aren't mutually exclusive.
Honestly, this sounds like some crap spread by people who can't bring themselves to believe that non-Western "primitive people" could have highly developed skills.
1. https://en.wikipedia.org/wiki/Polynesian_navigation
The system of navigation was recorded orally. People in living memory retained the knowledge. Anthropologists got it from them, built replica canoes, and then sailed around discovering that it worked. Captain Cook used Polynesian navigator who was well traveled.
Maybe the rafts are coming from Easter Island losing the technology and using rafts instead.
I can sort of see that in a mind's eye, with rings of waves spreading as they bounce off obstacles in water. But that's bird's view of a miniature -- and seeing that from the surface would be a very different story.
It’s amazing they can read such nuance.
> 1. A natural or inherent impulse or behaviour.
> 2. An intuitive reaction not based on rational conscious thought.
The man steered the boat with one had in the water, until he announced we were near a shoal of mackerel - he said he could feel the grease of their bodies in the sea water.
We threw some lines in and sure enough we caught plenty of fish for dinner.
They could see from the patterns in the waves, thousands of miles away, where the land masses were.
>insert joke about training neural networks
(If anyone knows the original quotes, please post below.)
"A depth of only 25 fathoms is quite enough to give some surface indications: coloration, wave phenomena, perhaps fauna. Is this the explanation of the ghost island? Some lost traveller, perhaps en route from Yap to Guam, seeing and remembering these phenomena, later reifying them as an inhabited land? Or is it possible that a real island once existed here, as the Carolinians say? [...] Any Carolinian navigator worthy of the name can give a whole set of radiating courses under all the navigation stars from every island of the Carolines, not just from Kaafiror. [... N]avigators do learn them, together with the courses from real islands, and they make no distinction among them. It is perhaps not altogether in the realm of fantasy to speculate that the curriculum of the schools of navigation was established in a time when Kaafifor was more than a discolored patch of water."
[1] https://micronesica.org/sites/default/files/the_ghost_island...
The boundary between "imagining" and "visualizing" is somewhere between these two experiences but conceptually they are not that different.
I'm not sure how the concept of tracking virtual islands over the horizon really helps. The only thing I can think of is maybe the idea of it encourages the navigator to stay focused on estimating the passage of proxy points on the far horizon based on whatever composite of wind, current and swell signs they are intuiting from. While still quite variable, I assume gauging distance estimates on the far horizon is better than the alternative of trying to estimate distance traveled from the immediate surroundings of the craft (which are only useful for estimating velocity).
You can practice navagating against physical islands over the horizon, and when you're good at that, you've mostly gotten good at dead reckoning against a real reference point; of course, with corrections from the islands influence. Having a community shared archipelego of virtual islands lets you focus your dead reckoning skills on a point while offering a vocabulary of distance and reducing travel times between waypoints.
Go 1000 miles in this direction seems a lot harder for me to follow over many days than go X miles to A, then Y miles to B, then Z miles to C. Even if A and B aren't real. If I treat them as very small islands that will be over the horizon, no big deal that they don't influence the environment, they're small; but I can't really use them to course correct, my reckoning needs to be good.
That is with experience you can select something on the ocean far away that you can track (kelp, etc), with experience and focus to accurately track it and take into account its own movement. Then the method is obviously only promoted by those who successfully survived using it, as the potential nay-sayers who used it and failed are no longer there to give a counter-point. Finally, those who did not use it successfully are probably characterized as "not good navigators", in circular logic.
(Bet they didn't think at the time that techies of the future would be watching it in 2024 on the ubiquitous global data network.)
But did they imagine that many techies a few decades in the future would be interested in the home video they were making?
Look for clips of "Star Wars" in action.
The 70's was anni interesting time. Atari was employing dynamic vector displays capable of real time motion.
Tektronix was using vectors with their storage CRT tech. Basically the vectors got painted onto the tube phosphors, thus displaying the image without the need to refresh.
4k resolution (vector coordinate space) ended up being a thing!
https://youtu.be/f8I8TtK_6sw?si=LQ1sZK6jt0QKhMNX
It seems that Etak was to navigation systems what Jodorowsky's Dune was to 1980s sci-fi: a trail blazing endeavour that was wild and wildly innovative, did not fulfill its intended mission but rather set up an entire field for subsequent success.
Also: the design must have included several masterpieces when considering the state of tech in the 1980s: even seeking to the right point on the map cassette is an untrivially hard problem.
What I wanted to point out is that both projects produced massive amounts of reusable knowledge, and that knowledge set up the stage for a whole field of influential and exciting derivative works.
You seem to be responding to another comment? Or are putting words in the commenters’ mouth?
Specific examples from the documentary were Star Wars and Alien. The latter of which included Dan O’Bannon (screenwriter) and H.R. Giger (effects+concept artist), reprising their roles as staff on Jodorowsky’s Dune.
Highly recommend the documentary by the way. I actually just saw it a couple weeks ago, so the comparison made a lot of sense to me.
Edit to add: I’d still unquestionably put Star Wars as the one that “set up an entire field for success”. Not just because it was the one that shipped and was a massive hit, but because it synthesised many elements besides Dune. I’ll watch that documentary with an open mind though.
> When I worked on the Apple Maps team, 12 of my colleagues were Etak alumni.
What a legacy! It's gratifying to hear of these long-dead companies/products with incredible engineers who are still out there slinging code with the best of them.
I'd think that it's everyone else at Apple who's slinging code with the best of them.
Interestingly, one of Etak co-founders was Nolan Bushnell, and he was Atari co-founder that hired Steve Jobs (or specifically Allan Alcorn) in his only full-time job prior to Apple, pardon the pun.
The tech behind mapblast also powered the first version of Yahoo Maps, which was a pretty big deal at the time (this was before Google Maps eventually came in and overshadowed everyone else in that market).
I was in my early 20s and working for one of my first tech startups at the time but Vicinity was primarily made up of graybeards who had previously worked for Etak (who was the primary map data provider for the online mapping system we made, so they had a lot of experience with it) and many of those people were also ex-Atari prior to Etak.
No way did they invent this. Not even close!
This is called map matching. It predates Etak by at least 20 years, if not more.
This paper was published a decade before which does exactly this: Lezniak TW, Lewis RW, Mcmillen RA. A dead reckoning/map correlation system for automatic vehicle tracking. IEEE Transactions on Vehicular Technology. 1977 Feb;26(1):47-60.
The government was building out this technology in the 50s, here's a RAND report about that. https://www.secretsdeclassified.af.mil/Portals/67/documents/...
I suspect there are even earlier examples.
Still impressive to get it into a consumer-sized (and almost consumer-priced) box.
I'm not disputing your assertion, but perhaps I'm a little more charitable in thinking they could have independently invented the same thing and believed they were the first since the one you mentioned was apparently classified.
Edit: Funnily enough, searching on google for this paper, your comment is the second result.
May be it was easier to market only for rich people who wish to show off then? Since the fall of Vertu no tech companies seem to address specifically this segment. Or may be people just were more optimistic about tech?
I'd rather think that there is more money frivolously spent today; in the S.F. Bay Area, much more.
> May be it was easier to market only for rich people who wish to show off then?
I'm quite perplexed about how your perception can be so very different from mine. How many people own a Tesla in your neighborhood?
This device however, I would have thought, would have been marketed chiefly to professionals. Traveling salesmen, doctors (who then still made house calls), service technicians etc. .
I'd pay for a CRT Waze skin and the option to turn off turn-by-turn navigation.
Later on they talk about "heads up" map digitizing, did this mean the map rotated as the operator digitized the street? Seems quite unwieldy (and how did the poor PC rotate raster graphics?)
For the in car map, a vector CRT and vector data makes rotation reasonable. Much less hard than rotating rasterized scanned images on a PC with no rotation acceleration.
digitizing blind at that point would mostly be used because it was cheaper than getting a graphics capable computer.
using a projector instead of a screen would be enough to devise a system where the digital image is shown on top of a printed map. so when they came up with that idea they probably already had most of the pieces they need to make it work.
Interesting side-fact: we used time-of-flight with beepers placed in the cars and ordinary trigonometry to increase accuracy. Worked like a charm.
Anyway I really need such articles from time to time not to lose my faith in humanity.
"The prototype driver navigation system was unveiled in Hildesheim on June 21, 1983, and it proved groundbreaking: EVA was the first ever experimental autonomous navigation system."
https://www.bosch.com/stories/eva-first-navigation-system-fr...
(no map, but display and address to address with route finding)
Both (Etak and TravelPilot IDS) seems to use kind of a vector display. Does someone know if this is for better resolution or better contrast, or both or if there is another reason?
https://www.bosch-presse.de/pressportal/de/en/navigation-sys...
Reminds me at one point in time I owned a Vectrex, which had much cleaner lines than any other console, even much better than my later, much more expensive Amiga or (early) PCs.
"This Vectrex does things I never thought possible" https://www.youtube.com/watch?v=8Dv15YRAmzM
Only got that feeling back with Retina displays.
This can occasionally, in rare situations, be a problem, if you have frontage roads very close to a highway, they can sometimes get confused about which road you are actually on.
If Waze instructs me to take the exit, then it assumes I'm on the service drive, even if GPS says I'm still on the freeway. And vice-versa, more problematically -- if I impulsively take the exit, it assumes I'm still on the freeway even if GPS clearly shows I'm on the service drive.
(I can confirm this by running a spare laptop with a USB GPS as a logger, while my iPhone runs Waze. Overlay the GPX on a map later and it's super obvious whether I took the exit or not, but either the Apple location provider or Waze staunchly ignores reality in favor of obsessive road snapping.)
Where this gets stupid is, if there's a traffic jam on the freeway and I dip onto the exit to avoid it, now Waze sees me flowing freely down the service drive, assumes that it's the freeway that's flowing freely, and disbelieves other users who report traffic there. Even as the service drive curves and diverges and I follow the curve, it doesn't retroactively say "Oh jeez, he must be on the service drive after all, adjust the previous data to apply to the service drive and not to the freeway!". So the bad data continues to corrupt the traffic picture and encourage other users to get stuck in traffic they can't report.
i struggle to imagine how did the tape handle it.
> Compact tape drive. which uses 1/4-in. magnetic tape cassettes operating at 80 ips, each containing every street and specific address, for an area about twice that of an ordinary paper street map, as well as overviews of major state and regional roads, and national interstates (installed under the vehicle dashboard or in the glove compartment).
So it seems that, if it's not 80 inches per second, then the confusion dates at least back to 1985!
another source <https://rosap.ntl.bts.gov/view/dot/2332/dot_2332_DS1.pdf> also reports:
> TAPE DRIVE
> 5” x 2 3/4” x 3 3/4" 80 ips
Tape shown in the video demonstration of Etak was ordinary 1/8 Philips 4 x 2 1/2 x 1/2.
>another source <https://rosap.ntl.bts.gov/view/dot/2332/dot_2332_DS1.pdf>
Dimensions are for a Tape Drive so that matches Philips cassette. Speed is 40x normal, but reading https://www.theregister.com/Print/2013/08/30/50_years_of_the...
"The actual production of Musicassettes was done on machines running 32 times faster than normal playback. Cassette tape would be reeled over four heads recording what would be both sides at once at 60 IPS. The master tape that was source of the original music had been recorded at 7.5 IPS and this would also run 32 times faster, clocking up a playback speed of 240 IPS for duplication purposes."
"This super-fast tape transport also required the circuitry to follow suit. So instead of the bias frequency being around 80kHz, it was now 2.4MHz; the amplifiers also needed to work over a frequency range of 200kHz to 500kHz."
Commodore 64 with best tape Turbo is able to store slightly above 1MB per cassette. Japan already had floppy drives capable of storing over 1MB in 1983, but it looks like Etak needed more, with this 40 times faster tape drive delivering:
"local map data base stored on a 3.5-MByte tape cassette."
I would love to learn more details about this drive. Modulation used? Number of tracks? Format? Magnetic flux dump of one of the cassettes would be a lovely puzzle to decode.
I remember installation wasn't trivial. It needed a lot of futzing with the car. I remember the wheel rotation sensors, and they are briefly mentioned in the article.
Smazing that nowadays all this stuff is solid state and in your pocket.