Slashdot Mirror
Dead Reckoning For Your Car Eliminates GPS Dead Zones
cartechboy writes "We've all been there. You're relying on your vehicle's built-in navigation system to get to that meeting downtown, but then suddenly the car loses the satellite signal due to the concrete skyscraper canyon you're in--and you're about to be late. Swiss semiconductor manufacturer U-Blox thinks it has the solution with 3D Automotive Dead Reckoning, or 3D ADR for short. It's a new navigation chip that uses the vehicle's built-in sensors to track speed, horizontal movement, and elevation. The 3D ADR system measures movement in three dimensions, letting the navigation system can keep track of the vehicle's location even when it loses its connection to GPS satellites. Imagine never having to see your navigation screen saying connection lost again. In an age where our phones have accelerometers and compasses, it's amazing your car is still trying to catch up, right?"
Seriously this isn't new. Good in-car nav systems have had dead reckoning based on wheel position + speed for ages.
Actually I think it's the opposite, it's only being in a car that makes dead reckoning with any kind of accuracy feasible. A car is a reasonably large and stable platform, which already has good speed information, and can have some accelerometer-type information added relatively easily. A smartphone does have an accelerometer, but the data is far too noisy to do reasonable dead reckoning, because in addition to the macro movements (someone walking, biking, or driving down a street) there area bunch of micro-movements that produce high local acceleration (putting the phone in/out of pockets, taking steps while the phone's in your pocket, etc.). It makes for a much more complex dead-reckoning problem, because instead of just tracking broad movements (car goes 10m this way) you have to resolve a ton of tiny movements (phone was moved 0.3m into pocket, then rapidly accelerated 0.1m left due to owner being jostled on the subway, etc.), which tend to pile too much accelerometer noise on top of the broader movements that you really want to track.
In short, taking a known starting position and keeping it updated via accelerometer data is a lot easier if your accelerometer is on a car, vs. on a handheld device.
10 PRINT CHR$(205.5+RND(1)); : GOTO 10
There was a system called ETAK (1983... see wikipedia). At the time they "Said" Etak was a Polynesian word that meant "the world moves" and that the technique came from the polynesian nevigation methods.
Nothing new to see here... Just repackaging masquerading as new
Beta, brought to you by the same people that fucked Sourceforge.
Fork or die.
Don't forget nntp://comp.misc on Usenet
(although I doubt there are any modern browsers that support the nntp:// protocol* anymore so you will have to join mainly by downloading a free usenet client, sign up to eternal-september.org and adding comp.misc to your subscribed newsgroup list)
Still, if people are willing to use old school IRC then Usenet isn't much worse ;-)
* similar to PIN number
My 7-year-old Prius tracks my movements in tunnels, inside a garage with no GPS, etc. It indicates my changing heading and position as I back out of a parking space, etc.
Beta is more than cosmetics or aesthetics. The new design ruins the one thing that makes /. what it is -- the commenting system. I only come here for the comments, not the 2-day old articles nor the erroneous summaries.
I do not see the changes of Beta as improvements. What is wrong with Slashdot that demands breaking its foundations? This is not change for the sake of change, but, as others have commented, an attempt to monetize /. at any any cost, and its users be damned.
Our complaints have fallen on deaf ears, and will continue to do so. Dice intends to dispose of Classic in favor of Beta, whether we like it or not. Do you know how to tell whether an executive really cares about feedback? If her CV doesn't proclaim the following "successes":
Proven track record innovating and improving iconic websites (CNET.com, Dice.com, Slashdot.org, Sourceforge.net) while protecting their voice and brand integrity
seriously, my old bimmer's on-board navigation already did this 15 years ago ! other than that, I've used u-blox in several embedded designs, and they are by far the most fun GPS unit to play with. They have some great pc software for it too. And they've had this functionality for quite some time now. oh yeah, and boo to beta..
The title is misleading. They use accelerators and vehicle speed to navigate. It aint dead reckoning
Why 3D? Because GPS provides your elevation and uses that as well as your XY coordinates to determine where you are, trip time, etc.
Contrary to the assertion in the story, I've never been there.
Don't know who well it works, but there was a demo by Apple's iOS developers where they combine GPS and WiFi. In towns with large buildings and awful GPS reception you will usually have tons of WiFi signals around, so at least in principle it should be possible to improve navigation using both.
Believe me, when your GPS tells you to go straight ahead because that's where the exit is but you're still on the 6th floor, yes, you definitely do!
We used to have a Bill of Rights. Now, with the rights gone, all we have left is the bill.
Yeah sure, we need several million Slashdot clones ... ;-)
The Tao of math: The numbers you can count are not the real numbers.
A few months ago I left work to run some errands and stuck my Android phone in its car charging dock (which automatically activated my preferred nav program). Six or seven miles down the road I noticed the icon representing my car was different than normal and my location was about a half-block off my actual position. At the next stoplight I checked and discovered my GPS was turned off. My phone had reasonably calculated my position through several turns and stops using only the accelerometer (dead-reckoning). At the time I was impressed.
Beta sux! Join the Slashcott! http://hardware.slashdot.org/comments.pl?sid=4760465&cid=46173047
Who would use GPS to navigate inside a parking garage?
"Those who consume the bulk of goods are those who make them. We must never forget this secret of our prosperity."
My old cheap Samsung tracks via accel. and compass most of the time, because the GPS is so poor. As long as it can get a GPS fix every few minutes it covers up for the crappy GPS antenna quite well. IIRC the 'Tomtom' that I used to use at work would do the same thing, only of course it didnt need the capability near as often, but you could drive through a long tunnel with it and still show actual position until the signal re-established at the other end.
Presumably there is something new here, but the basic concept certainly is not.
=-=-=-=-=-=-=-=-=-=-=-=-=-=-
Friends don't let friends enable ecmascript.
People that need to use a GPS inside a parking garage, for one.
rewriting history since 2109
Being a Sprint customer?
Use of accelerometers is only to reduce the error. Unfortunately, accelerometers can be wrong - due to rotation, deceleration, and acceleration when there is no feedback on WHAT is causing the readings
Cars don't fly. As long as you have traction, measuring distance shouldn't be a problem (especially if you supplement wheel axle sensors with an optical device similar to modern computer mice, to compensate for the tire pressure, and integrate the data). Unlike with airplanes and ICBMs, you shouldn't need precise acceleration from the accelerometers, vehicle orientation along the three axes is what you need. Direct distance measurements obviate the need for double integration (but are impossible in subs, airplanes, and spacecrafts, unlike in cars).
Ezekiel 23:20
GPS elevation measurements are notoriously problematic unless you have survey-class equipment. (I also wonder how many map application programmers can actually work with WGS84 elevation data.)
Ezekiel 23:20
A self-driving car? (In conjunction with other sensors, I assume.)
Ezekiel 23:20
The automotive nav system I work on has 2D dead reckoning, relying on GPS for altitude. 3D dead reckoning is news to me, and I suspect is the intended newsworthy bit here. When GPS fails and the digital terrain model doesn't account for urban landscapes like parking garages, both above and below ground, and tunnels, positioning is calculated in software. It's an expensive, imperfect process I imagine can at least be offloaded to sensors, if not improved, as possibly done in this chip. That would certainly free up the radio to provide more accurate positioning and guidance quicker.
Yep, my 2001 Acura has it too. They use wheel sensor data for speed and a gyro compass for azimuth. It works very well.
A GPS which receives speed, wheel position and reverse light data from the car does dead reckoning. I watched a friend's car do it just this weekend as we drove into an underground parking garage where you get no GPS reception.
Dead reckoning for car NAVs has been around forever, it actually predates GPS. The first in-car NAV systems by Etak were made using only maps and dead reckoning because GPS didn't exist yet. It also predated affordable LCD panels.
Before accelerometers and gyros were cheap Garmin even made add-on GPSes for cars which required installation so the NAV could get speedometer data and reverse light data so it could dead reckon your position.
http://en.wikipedia.org/wiki/E...
http://lkml.org/lkml/2005/8/20/95
All car navigation systems pretty much required this when the GPS system was still hobbled by the ~100m uncertainty caused by Selective Availability. (Ended by Clinton in May 2000).
The implementation is actually quite trivial: One sensor on each front wheel gives you two revolution counters (odometers).
Distance traveled is proportional to the sum of the two counters, while the difference in counts is proportional to how much you have turned.
As long as you have GPS reception you can use that to calibrate the odometers, so that differences in tire type & pressure is automatically compensated for.
Using a barometer you can do the same for altitude, automatically compensating for changes in local air pressure.
Terje
"almost all programming can be viewed as an exercise in caching"
Who would use GPS to navigate inside a parking garage?
Not to navigate, hopefully. But in a large parking garage that you don't know well, it would be useful to be able to use GPS to remember where your car is and find it again.
You don't even need to go to all that trouble. That kind of accuracy simply isn't necessary in a car as long as it has GPS. Dead-reckoning, in an in-car navigation system, is only needed to make up for the inaccuracy with GPS, which is mainly because you sometimes lose the signals (inside a parking garage, around too many tall buildings, etc.). You don't normally lose GPS signals for a very long amount of time, only short periods. So DR is just a supplement, and the vehicle's odometer signal alone is enough for supplementation, and an electronic compass signal is even better because this gives you direction, in case you turn the car while your GPS signal is missing. You're not looking for accuracy to the inch or even the foot, 50-foot accuracy is good enough in this application, so compensating for tire pressure is entirely unnecessary; they don't do that for the vehicle's odometer after all. You just need the system to warn you in enough time for you to make the next turn.
Nuclear submarines use them too! :-)
Lisias@Earth.SolarSystem.OrionArm.MilkyWay.Local.Virgo.Universe.org
Without good DR you don't have to lose a GPS signal in a large city for a nav system's positioning to utterly fail. An Urban Canyoning effect can render GPS unreliable despite a strong signal. I've seen GPS in a mile long urban canyon position a vehicle in and across a parallel river in the middle of the canyon. The ultimate solution when this behavior is detected is to rely solely on DR, which in practice can go days of driving before being significantly inaccurate.
Why 3D? Because some places have lots of elevated highways (as well as parallel roads underneath them).
But mostly because GPS works in 3D, and re-acquiring the signal is faster the more accurate the 3D estimated location is.
It's probably too late to turn back the clock on this, but it's actually spelled "ded" reckoning. Short for "DEDuced".
Surely you can use both, even if your DR data is not so great; just use the DR data to verify the GPS data makes sense. If the GPS signals say the vehicle has suddenly teleported into a river, check that against the DR. If the DR data says you're still driving straight, then disregard the GPS data until it agrees with the DR data.
I can only see one possible outcome of that. -- "Uh, where was my car again?" - (checks GPS) - "Oh, right, it's still right under my butt, duh."
CLI paste? paste.pr0.tips!
"accelerometers can be wrong - due to rotation"
That's why you also have a 3-d gyro.
"deceleration"
Um. Accelerometers measure both acceleration and deceleration...
"and acceleration when there is no feedback"
Like when the user has dropped their phone and broken the accelerometer?
What about in the maze of tunnels that run under Sydney?
When I worked for Navman we developed some of the first consumer in car navigation systems. We looked at this technology around 2002. Nothing new here...
I know I can upset my ABS system, bringing on the the error light, by doing burnouts and handbrake slides. Burnouts give false readings on the from wheels and handbrake slides give false reading on the rear wheels.
Would play havoc with such systems but to be fair I don't normally do these while navigating.
Dead reckoning technology is actually very old. It has been used to guide missiles, submarines, and of course cars for decades even before the GPS was invented. It is the technology used by sailors before they had GPS as well. The idea is simple and complex at the same time: use some specific known reference, guess what's happening in the absence of reference, and recalibrate once a new reference becomes available again . References can be the sun, stars, towns, or GPS itself.
In car dead reckoning, in contrary to what the article says, you typically don't use acceleration sensors. You typically use vehicle speed and yaw rate sensor. This gives you enough information to determine whether you've turned, and where you are along the road. If you can safely assume that you are following the road on your digital map, this is actually quite accurate. It becomes tricky if you are airborne and free to fly around, but also possible.
The nice thing about GPS is that the kalman filter used to compute your position and velocity can be easily extended to include additional sensors such as yaw rate and speed, available on any modern vehicles CAN bus. The only trick is to have the navigation system hooked into the vehicle, and this is one of the main advantage of built-in systems (the other being driver assistance functions taking advantage of map data for enhanced functionality)
I think there were some navigation systems manufactures trying to achieve similar results by adding accelerometers to the receivers. Since people usually use these devices to follow a guided route, a yaw rate sensor to detect turns is not essential, and detecting stop conditions in urban canyons or tunnels can be detected via accelerometers.
The possibilities are endless and they have been used forever in the navigation industry. The article is extremely misleading by claiming that this is new, or hasn't been done before. Nevertheless a cool technology.
You don't even need to go to all that trouble. That kind of accuracy simply isn't necessary in a car as long as it has GPS.
If you define "trouble" as the cost of hardware, I'd argue that I've outlined a route that is as trouble-free as possible, since it doesn't require anything fancy (in terms of manufacturing costs - a high-precision inertial platform would be an example of "fancy" in this sense).
That kind of accuracy simply isn't necessary in a car as long as it has GPS.
I was actually looking forward, towards automated cars. But then again, those will have situational awareness requirements so high that deriving the motion data from CV might be a necessary option anyway (at least in urban settings - in open landscapes, that might fail, but then again, there would be a good GPS signal to rely on instead).
Ezekiel 23:20
Ooo...! You! I mean... What? Oooh! (Or rather, What about internal consistency?)
Yes, both are always used, but what I described above was a corner case where GPS is unreliable, with a good signal too, for a long stretch so DR has to be solely relied upon, unlike what you suggested earlier that positioning be biased toward GPS. In that example, GPS had preference and was positioning the vehicle in the river. Because situations like this occur regularly in large cities, in addition to a lost GPS signal, positioning is more accurate when biased toward DR.
It seems like the algorithm could be tweaked a little so that if the GPS is showing the car in a river, a forest, a building, etc. instead of driving on a road as DR indicates, then DR should be favored. After all, it simply doesn't make any sense for the vehicle to be in a river. Or, if you're taking an exit ramp, and the GPS signals show the vehicle veering off into the trees while DR just shows the vehicle is proceeding around the curve at a normal rate, then it should be pretty obvious that the vehicle has not, in fact, gone off-road and is driving through metal barriers and trees at 40mph.
Errr, no.
What is the usage case again?
You're in a concrete canyon downtown. So what the fuck are you doing driving? That's what taxis and tube trains and buses are for. Cars in the city centre are almost always a guarantee for frustration and delay.
Besides, if it's your own city, how on earth can you exist without knowing it at least as well as the Sat Nav, and also knowing the pedestrian-only short cuts that go against the flow of one-way streets, across parks and directly under 14-way road junctions designed to get people from somewhere where you didn't start, going to somewhere you're not going to.
Do people not use their brains these days?
Birds are not dinosaur descendants;birds are dinosaurs, for all useful meanings of "birds", "are" and "dinosaurs"
One word, Etak.
Dead reckoning automotive navigation predates GPS and predates what would have become Geostar. Sure, the hardware is cheaper now, and algorithms might be better, but there's nothing new about this concept. Welcomed? Sure. But don't pretend it's a new idea.
Car electronics are perpetually shit, out of date, and overpriced. Put the feature on my phone I don't care if it won't be as accurate.