Re: Accelerometer brainstorming
Hello, On Tue, Apr 1, 2008 at 5:49 PM, Alexey Feldgendler <[EMAIL PROTECTED]> wrote: > > Are roundabouts in tunnels an April fool? Never encountered one. No, roundabouts in tunnels are real - we even have one or two such tunnels here in Norway. -- Regards, Torfinn Ingolfsen Norway ___ Openmoko community mailing list community@lists.openmoko.org http://lists.openmoko.org/mailman/listinfo/community
Re: Accelerometer brainstorming
I know from my experience with 3D accelerometers that there is a lot of drifting. We would also need a filter to take away vibrations. But I also know that some guys at my Uni. managed to keep a model helicopter at the same position in the air by only using input from the accelerometers and the result was amazing:) (I will try to figure out if the code is still available.) -- Please don't send me Word or PowerPoint attachments. See http://www.gnu.org/philosophy/no-word-attachments.html Join the FSF as an Associate Member at: http://www.fsf.org/register_form?referrer=5774> ___ Openmoko community mailing list community@lists.openmoko.org http://lists.openmoko.org/mailman/listinfo/community
Re: Accelerometer brainstorming
Flemming Richter Mikkelsen wrote: What do we need the CAN interface for? We already know the speed before we enter the tunnel, and if the neo is in a car holder in a stable position, calibrated with some software, it knows from the accelerometers if we are driving strait ahead or making a turn and also if we are accelerating. With a little bit of mathematics, this can turn out to be very precise. Precise. I don't know... TomTom had to use some mighty black magic to make the sensor drivers somewhat realtime aware. AFAIK Just my 2 Eurocents Tilman ___ Openmoko community mailing list community@lists.openmoko.org http://lists.openmoko.org/mailman/listinfo/community
Re: Accelerometer brainstorming
Alexy is correct. Due to the integrative nature of accelerometers, the longer the time, the more error associated with any result. On Wed, Apr 2, 2008 at 12:09 PM, Alexey Feldgendler <[EMAIL PROTECTED]> wrote: > On Wed, 02 Apr 2008 17:52:56 +0200, Flemming Richter Mikkelsen < > [EMAIL PROTECTED]> wrote: > > What do we need the CAN interface for? > > We already know the speed before we enter the tunnel, and if the neo > > is in a car holder in a stable position, calibrated with some > > software, it knows from the accelerometers if we are driving strait > > ahead or making a turn and also if we are accelerating. With a little > > bit of mathematics, this can turn out to be very precise. > > > > As you're driving through a long tunnel (the longest one in Norway being > 24.5 km), error accumulation will deteriorate the precision to the point of > rendering the data useless. > > > -- > Alexey Feldgendler <[EMAIL PROTECTED]> > [ICQ: 115226275] http://feldgendler.livejournal.com > > ___ > Openmoko community mailing list > community@lists.openmoko.org > http://lists.openmoko.org/mailman/listinfo/community > ___ Openmoko community mailing list community@lists.openmoko.org http://lists.openmoko.org/mailman/listinfo/community
RE: Accelerometer brainstorming
I think that generally mems devices have proven to work badly for this sort of thing. Because of random drift and other errors. It would need to assume that the car is on a path, the road, and attempt to infer where on the path it is. It wouldn't likely work with just the acell data, and it would take some finely tuned software and then some to get even an aproximation. Of course, it might be ok for the typical case of periodic short term obscured gps signal. Any distance or turns and it's liekly to diverge pretty fast. Matt -Original Message- From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] On Behalf Of Flemming Richter Mikkelsen Sent: Wednesday, April 02, 2008 11:53 AM To: List for Openmoko community discussion Subject: Re: Accelerometer brainstorming On 4/1/08, Al Johnson <[EMAIL PROTECTED]> wrote: > On Tuesday 01 April 2008, Flemming Richter Mikkelsen wrote: > > On 4/1/08, Alexey Feldgendler <[EMAIL PROTECTED]> wrote: > > > In Norway, where I live, some tunnels are very long and even sometimes > > > contain internal enter and exit lanes, so there can be more than one way > > > to go. My Garmin navigator tries to estimate where I am basing on the > > > speed I had when entering the tunnel, so sometimes it notifies me of an > > > internal exit I have to take too late. > > > > Correct! We even have round abouts inside tunnels. If you take the wrong > > exit, > > you would have to drive for a long time to be able to turn and get back (on > > the > > highway). So the phone needs a system that is better than the one from > > Garmin. > > > > -=Flemming=- > > So you need a car cradle with a CAN interface - IIRC there have been USB-CAN > adaptors listed here before. This would give you access to the vehicle speed, > and perhaps other useful things - steering angle perhaps? Accelerometer input > for finding corners may be useful even if it's not accurate enough for > inertial navigation. The interesting part comes in combining the available > data to give the location estimate. > What do we need the CAN interface for? We already know the speed before we enter the tunnel, and if the neo is in a car holder in a stable position, calibrated with some software, it knows from the accelerometers if we are driving strait ahead or making a turn and also if we are accelerating. With a little bit of mathematics, this can turn out to be very precise. -- Please don't send me Word or PowerPoint attachments. See http://www.gnu.org/philosophy/no-word-attachments.html Join the FSF as an Associate Member at: http://www.fsf.org/register_form?referrer=5774> ___ Openmoko community mailing list community@lists.openmoko.org http://lists.openmoko.org/mailman/listinfo/community ___ Openmoko community mailing list community@lists.openmoko.org http://lists.openmoko.org/mailman/listinfo/community
Re: Accelerometer brainstorming
On Wed, 02 Apr 2008 17:52:56 +0200, Flemming Richter Mikkelsen <[EMAIL PROTECTED]> wrote: What do we need the CAN interface for? We already know the speed before we enter the tunnel, and if the neo is in a car holder in a stable position, calibrated with some software, it knows from the accelerometers if we are driving strait ahead or making a turn and also if we are accelerating. With a little bit of mathematics, this can turn out to be very precise. As you're driving through a long tunnel (the longest one in Norway being 24.5 km), error accumulation will deteriorate the precision to the point of rendering the data useless. -- Alexey Feldgendler <[EMAIL PROTECTED]> [ICQ: 115226275] http://feldgendler.livejournal.com ___ Openmoko community mailing list community@lists.openmoko.org http://lists.openmoko.org/mailman/listinfo/community
Re: Accelerometer brainstorming
> What do we need the CAN interface for? > We already know the speed before we enter the tunnel, and if the neo > is in a car holder in a stable position, calibrated with some > software, it knows from the accelerometers if we are driving strait > ahead or making a turn and also if we are accelerating. With a little > bit of mathematics, this can turn out to be very precise. Yes, with the phone fixed to the car, and with initial position and velocity, we can calculate the position (and velocity) in the future. The only problem is numerical errors from the method and imprecision from the initial values. So having the velocity of the car can really help, to correct those errors, but it's not necessary. It could be a nice demo app for the GSoC work on the accelerometers. ___ Openmoko community mailing list community@lists.openmoko.org http://lists.openmoko.org/mailman/listinfo/community
Re: Accelerometer brainstorming
On 4/1/08, Al Johnson <[EMAIL PROTECTED]> wrote: > On Tuesday 01 April 2008, Flemming Richter Mikkelsen wrote: > > On 4/1/08, Alexey Feldgendler <[EMAIL PROTECTED]> wrote: > > > In Norway, where I live, some tunnels are very long and even sometimes > > > contain internal enter and exit lanes, so there can be more than one way > > > to go. My Garmin navigator tries to estimate where I am basing on the > > > speed I had when entering the tunnel, so sometimes it notifies me of an > > > internal exit I have to take too late. > > > > Correct! We even have round abouts inside tunnels. If you take the wrong > > exit, > > you would have to drive for a long time to be able to turn and get back (on > > the > > highway). So the phone needs a system that is better than the one from > > Garmin. > > > > -=Flemming=- > > So you need a car cradle with a CAN interface - IIRC there have been USB-CAN > adaptors listed here before. This would give you access to the vehicle speed, > and perhaps other useful things - steering angle perhaps? Accelerometer input > for finding corners may be useful even if it's not accurate enough for > inertial navigation. The interesting part comes in combining the available > data to give the location estimate. > What do we need the CAN interface for? We already know the speed before we enter the tunnel, and if the neo is in a car holder in a stable position, calibrated with some software, it knows from the accelerometers if we are driving strait ahead or making a turn and also if we are accelerating. With a little bit of mathematics, this can turn out to be very precise. -- Please don't send me Word or PowerPoint attachments. See http://www.gnu.org/philosophy/no-word-attachments.html Join the FSF as an Associate Member at: http://www.fsf.org/register_form?referrer=5774> ___ Openmoko community mailing list community@lists.openmoko.org http://lists.openmoko.org/mailman/listinfo/community
Re: Accelerometer brainstorming
I mean we could also use the open moko to chiptune the engine of our car or to control the ABS/ESP. or we could make it run the car so that nobody needs to pay attension to the road but is that really nessesary or a good idea? I mean it is not a super computer having a really large scale Cpu. and besides that I curently don't have any GPS I wonder how I did find my way through mess till now. maybe because there are in some rare cases big signs and yeah those are not even digital. The point is there might be some rare cases where we don't absolutely need a digital helper telling us how to handle a situation. We might have reason enough by ourself. ___ Openmoko community mailing list community@lists.openmoko.org http://lists.openmoko.org/mailman/listinfo/community
Re: Accelerometer brainstorming
On Tuesday 01 April 2008, Flemming Richter Mikkelsen wrote: > On 4/1/08, Alexey Feldgendler <[EMAIL PROTECTED]> wrote: > > In Norway, where I live, some tunnels are very long and even sometimes > > contain internal enter and exit lanes, so there can be more than one way > > to go. My Garmin navigator tries to estimate where I am basing on the > > speed I had when entering the tunnel, so sometimes it notifies me of an > > internal exit I have to take too late. > > Correct! We even have round abouts inside tunnels. If you take the wrong > exit, > you would have to drive for a long time to be able to turn and get back (on > the > highway). So the phone needs a system that is better than the one from > Garmin. > > -=Flemming=- So you need a car cradle with a CAN interface - IIRC there have been USB-CAN adaptors listed here before. This would give you access to the vehicle speed, and perhaps other useful things - steering angle perhaps? Accelerometer input for finding corners may be useful even if it's not accurate enough for inertial navigation. The interesting part comes in combining the available data to give the location estimate. ___ Openmoko community mailing list community@lists.openmoko.org http://lists.openmoko.org/mailman/listinfo/community
Re: Accelerometer brainstorming
On Tue, 01 Apr 2008 17:10:20 +0200, Flemming Richter Mikkelsen <[EMAIL PROTECTED]> wrote: In Norway, where I live, some tunnels are very long and even sometimes contain internal enter and exit lanes, so there can be more than one way to go. My Garmin navigator tries to estimate where I am basing on the speed I had when entering the tunnel, so sometimes it notifies me of an internal exit I have to take too late. Correct! We even have round abouts inside tunnels. If you take the wrong exit, you would have to drive for a long time to be able to turn and get back (on the highway). So the phone needs a system that is better than the one from Garmin. Are roundabouts in tunnels an April fool? Never encountered one. -- Alexey Feldgendler <[EMAIL PROTECTED]> [ICQ: 115226275] http://feldgendler.livejournal.com ___ Openmoko community mailing list community@lists.openmoko.org http://lists.openmoko.org/mailman/listinfo/community
Re: Accelerometer brainstorming
On 4/1/08, Alexey Feldgendler <[EMAIL PROTECTED]> wrote: > > In Norway, where I live, some tunnels are very long and even sometimes > contain internal enter and exit lanes, so there can be more than one way > to go. My Garmin navigator tries to estimate where I am basing on the > speed I had when entering the tunnel, so sometimes it notifies me of an > internal exit I have to take too late. > Correct! We even have round abouts inside tunnels. If you take the wrong exit, you would have to drive for a long time to be able to turn and get back (on the highway). So the phone needs a system that is better than the one from Garmin. -=Flemming=- -- Please don't send me Word or PowerPoint attachments. See http://www.gnu.org/philosophy/no-word-attachments.html Join the FSF as an Associate Member at: http://www.fsf.org/register_form?referrer=5774> ___ Openmoko community mailing list community@lists.openmoko.org http://lists.openmoko.org/mailman/listinfo/community
Re: Accelerometer brainstorming
On Tue, 01 Apr 2008 16:09:21 +0200, Christian Beier <[EMAIL PROTECTED]> wrote: Unfortunately you need accurate speed to be able to get usable data from accelerometers for that. That's why some GPS modules have an interface for tachometer. May be the history of GPS could be used to approximate the speed. Of course, it's not accurate but when you assume facts like: * you have already slow down your vehicle before entering the tunnel * you will never accelerate in the tunnel * max tunnel length of 3km the variance should not be so huge. At least it's easier than connect the tachometer. The other question is: do you really need the correct position in the tunnel? If you know the position when the signal get lost and know that you are in that special tunnel it should be enough. Of course, it makes more sens in urban "tunnels" between huge buildings. In Norway, where I live, some tunnels are very long and even sometimes contain internal enter and exit lanes, so there can be more than one way to go. My Garmin navigator tries to estimate where I am basing on the speed I had when entering the tunnel, so sometimes it notifies me of an internal exit I have to take too late. -- Alexey Feldgendler <[EMAIL PROTECTED]> [ICQ: 115226275] http://feldgendler.livejournal.com ___ Openmoko community mailing list community@lists.openmoko.org http://lists.openmoko.org/mailman/listinfo/community
Re: Accelerometer brainstorming
Am Fri, 28 Mar 2008 19:09:14 +0100 schrieb Peter Kraker <[EMAIL PROTECTED]>: > Unfortunately you need accurate speed to be able to get usable data > from accelerometers for that. That's why some GPS modules have an > interface for tachometer. May be the history of GPS could be used to approximate the speed. Of course, it's not accurate but when you assume facts like: * you have already slow down your vehicle before entering the tunnel * you will never accelerate in the tunnel * max tunnel length of 3km the variance should not be so huge. At least it's easier than connect the tachometer. The other question is: do you really need the correct position in the tunnel? If you know the position when the signal get lost and know that you are in that special tunnel it should be enough. Of course, it makes more sens in urban "tunnels" between huge buildings. Correct me if I'm wrong, don't know tunnels outside Europe. Christian ___ Openmoko community mailing list community@lists.openmoko.org http://lists.openmoko.org/mailman/listinfo/community
RE: Accelerometer brainstorming
Thanks. This package's guts might work as a starting point for gesture recognition. Prob some other fos out there sw does similar things. http://www.risujin.org/cellwriter From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] On Behalf Of Kyle Bassett Sent: Monday, March 31, 2008 1:25 PM To: List for Openmoko community discussion Subject: Re: Accelerometer brainstorming I think your examples are very possible. "Kind of reaching" would be something like trying to detect the type of jacket the person is wearing based on how much the fibers stretch during each step...or whether their jacket is open, by measuring the horizontal swinging motion. If the acceleration curve is very crisp, then the phone is prolly in their pant pocket. ;-) but I still say these things are not *impossible*, just require some very intuitive software. -Kyle On Fri, Mar 28, 2008 at 12:59 PM, Crane, Matthew <[EMAIL PROTECTED]> wrote: * I mentioned this in another thread, but if the device: - goes from 60km+ to 0 in short order, e.g. high g stop - while traveling horizontally - over a road - an on-screen alert/countdown is not stoped Then it's likely a vehicular accident so auto-call/sms for help with some kind of countdown to disable. Only really possible to do that with a phone + GPS + acell. * Sense when the owner is in a car by keying on the vibration signature and acceleration, possibly with gps data, and automatically bring up the assigned travel/map application. * Use acell data to charecterize person carrying the phone (as many couples out there will share phones, or give to children) and tie into user profile. * Detect when phone has dropped out of ones pocket, short fall by aprupt stop, beep three times loudly. (that's kind of reaching..) Matt -Original Message- From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] On Behalf Of Alexey Feldgendler Sent: Friday, March 28, 2008 9:22 AM To: List for OpenMoko community discussion Subject: Accelerometer brainstorming Just like my recent post about using GPS to enhance the phone, I'll try to explore the synergy opportunities for the accelerometers that Neo has. Once again, some of this might have already been discussed. * Intuitive mute: put the ringing phone on a flat horizontal surface (table) with its screen facing down to stop the ringing. Intuitive for the user and easy to detect with the accelerometers. Could also be used to hang up the current call. The gesture is very distinguishable and is hard to misinterpret because there aren't many situations when the user would normally put the phone on the table screen down. * Vibration control: when the phone is lying flat, either with its screen up or down, don't vibrate, so as to not produce the annoying loud noise. Ring instead, or do something else. Maybe make a weaker vibration, if the vibrator can do that. * Step counter: use the accelerometers to count steps when walking or running. Some people use dedicated devices for that. * GPS power saving: you can't move if you don't accelerate, so don't do power-expensive location detection until acceleration is detected. Once it's detected, start monitoring the location and keep doing it. Stop monitoring once the position stops changing for a while. -- Alexey Feldgendler <[EMAIL PROTECTED]> [ICQ: 115226275] http://feldgendler.livejournal.com ___ Openmoko community mailing list community@lists.openmoko.org http://lists.openmoko.org/mailman/listinfo/community ___ Openmoko community mailing list community@lists.openmoko.org http://lists.openmoko.org/mailman/listinfo/community ___ Openmoko community mailing list community@lists.openmoko.org http://lists.openmoko.org/mailman/listinfo/community
Re: Accelerometer brainstorming
I think your examples are very possible. "Kind of reaching" would be something like trying to detect the type of jacket the person is wearing based on how much the fibers stretch during each step...or whether their jacket is open, by measuring the horizontal swinging motion. If the acceleration curve is very crisp, then the phone is prolly in their pant pocket. ;-) but I still say these things are not *impossible*, just require some very intuitive software. -Kyle On Fri, Mar 28, 2008 at 12:59 PM, Crane, Matthew <[EMAIL PROTECTED]> wrote: > > * I mentioned this in another thread, but if the device: >- goes from 60km+ to 0 in short order, e.g. high g stop >- while traveling horizontally >- over a road >- an on-screen alert/countdown is not stoped > Then it's likely a vehicular accident so auto-call/sms for help with > some kind of countdown to disable. Only really possible to do that with > a phone + GPS + acell. > > * Sense when the owner is in a car by keying on the vibration signature > and acceleration, possibly with gps data, and automatically bring up the > assigned travel/map application. > > * Use acell data to charecterize person carrying the phone (as many > couples out there will share phones, or give to children) and tie into > user profile. > > * Detect when phone has dropped out of ones pocket, short fall by aprupt > stop, beep three times loudly. (that's kind of reaching..) > > Matt > > > -Original Message- > From: [EMAIL PROTECTED] > [mailto:[EMAIL PROTECTED] On Behalf Of Alexey > Feldgendler > Sent: Friday, March 28, 2008 9:22 AM > To: List for OpenMoko community discussion > Subject: Accelerometer brainstorming > > > Just like my recent post about using GPS to enhance the phone, I'll try > to > explore the synergy opportunities for the accelerometers that Neo has. > Once again, some of this might have already been discussed. > > * Intuitive mute: put the ringing phone on a flat horizontal surface > (table) with its screen facing down to stop the ringing. Intuitive for > the > user and easy to detect with the accelerometers. Could also be used to > hang up the current call. The gesture is very distinguishable and is > hard > to misinterpret because there aren't many situations when the user would > > normally put the phone on the table screen down. > > * Vibration control: when the phone is lying flat, either with its > screen > up or down, don't vibrate, so as to not produce the annoying loud noise. > > Ring instead, or do something else. Maybe make a weaker vibration, if > the > vibrator can do that. > > * Step counter: use the accelerometers to count steps when walking or > running. Some people use dedicated devices for that. > > * GPS power saving: you can't move if you don't accelerate, so don't do > > power-expensive location detection until acceleration is detected. Once > > it's detected, start monitoring the location and keep doing it. Stop > monitoring once the position stops changing for a while. > > > -- > Alexey Feldgendler <[EMAIL PROTECTED]> > [ICQ: 115226275] http://feldgendler.livejournal.com > > ___ > Openmoko community mailing list > community@lists.openmoko.org > http://lists.openmoko.org/mailman/listinfo/community > > ___ > Openmoko community mailing list > community@lists.openmoko.org > http://lists.openmoko.org/mailman/listinfo/community > ___ Openmoko community mailing list community@lists.openmoko.org http://lists.openmoko.org/mailman/listinfo/community
RE: Accelerometer brainstorming
I'm going to have a dig through citeseer for some ideas on representing accelerometer data in reasonably practical and time/scale/affine invariant form(s). I'd guess any libs/api would prefer to use fixed point if possible? Don't know what the chip resources are for fp math off hand. I notice this article on gait matching. I guess the best use of such a thing would likely be as the paper suggests, as a biometric authentication. Phone would stop working, without any obvious reason, if the gait didn't match. Certainly, with a stolen phone the first thing that happens to it would be it walks away. http://www.academypublisher.com/jcp/vol01/no07/jcp01075159.pdf Matt -Original Message- From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] On Behalf Of steve Sent: Saturday, March 29, 2008 9:45 PM To: 'List for Openmoko community discussion' Subject: RE: Accelerometer brainstorming Ya, I started down the API path with Michael Shiloh a while back, but we got diverted by other things. Hi matt, Lets talk about your ideas, we need to highlight the new hardware we added. Let me know what you are thinking -Original Message- From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] On Behalf Of Crane, Matthew Sent: Friday, March 28, 2008 11:08 AM To: List for Openmoko community discussion Subject: RE: Accelerometer brainstorming Yea, there would need to be a decent test and experiment app, with GUI, record data feature, and visualization of data. That would be needed likely before the ideas we've mentioned could be worked out. Need to develop a solid API and event notification system. That sort of thing prob exists for GPS, but should it be new/separate or tagged onto GPS for acellerometer? Matt -Original Message- From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] On Behalf Of Ben Burdette Sent: Friday, March 28, 2008 1:40 PM To: List for Openmoko community discussion Subject: Re: Accelerometer brainstorming Alexey Feldgendler wrote: > Just like my recent post about using GPS to enhance the phone, I'll > try to explore the synergy opportunities for the accelerometers that > Neo has. Once again, some of this might have already been discussed. I'd like to see an accelerometer graphing application, so you could monitor and graph g force levels. Maybe throw in a zeroing button to set a baseline orientation, then show deviation from that. Maybe not extraordinarily useful, but it would be cool to look at. You could also record the data for later hacking or comparison, like what was my max g-force last time I went skydiving compared to this time? The graphing and etc would be good for studying the feasability of other accelerometer uses, ie is a tap on the case a recognizable event? Speaking of that, I'd like a tap on the case to be configurable as a mute function during ringing. So if you forgot to mute your phone at the symphony and it starts ringing somewhere inside your backpack or jacket, a well placed kick would (probably) mute it. The value here is not having to hunt for the phone and find a button on it. I guess this falls under the larger category of gesture recognition, but this alone would be a real win for me. Another use would be as a 0-60 or quarter mile acceleration timer. Here's a link to a company that makes these: http://www.gtechpro.com/?gclid=CLf7wPmpsJICFRI0awod1CWyQQ I don't know if the accelerometer in the freerunner is accurate enough, but this is a low priced product, I'd think their accelerometers are not a great deal better, if any. ___ Openmoko community mailing list community@lists.openmoko.org http://lists.openmoko.org/mailman/listinfo/community ___ Openmoko community mailing list community@lists.openmoko.org http://lists.openmoko.org/mailman/listinfo/community ___ Openmoko community mailing list community@lists.openmoko.org http://lists.openmoko.org/mailman/listinfo/community ___ Openmoko community mailing list community@lists.openmoko.org http://lists.openmoko.org/mailman/listinfo/community
Re: Accelerometer brainstorming
On Mon, Mar 31, 2008 at 2:06 PM, Jay Vaughan <[EMAIL PROTECTED]> wrote: > > You folks are freaking me out. Outside of tracking people, any other > > ideas for why you'd want this? > > > reality-based games. > > pathfinding. > > picnics. > > parties in the desert. etc. > The cool factor? ___ Openmoko community mailing list community@lists.openmoko.org http://lists.openmoko.org/mailman/listinfo/community
Re: Accelerometer brainstorming
You folks are freaking me out. Outside of tracking people, any other ideas for why you'd want this? reality-based games. pathfinding. picnics. parties in the desert. etc. ; -- Jay Vaughan ___ Openmoko community mailing list community@lists.openmoko.org http://lists.openmoko.org/mailman/listinfo/community
Re: Accelerometer brainstorming
On Mon, Mar 31, 2008 at 6:45 AM, Gilles Casse <[EMAIL PROTECTED]> wrote: > On Lun 31 mars 2008 10:59, Kalle Happonen a écrit : > > > Alexey Feldgendler wrote: > >> > >> Do you think it's possible to use traits of a person's walk for > >> identification? Never heard about something like this. Interesting > >> idea, if it turns out implementable. > >> > > I remember seeing some research results about this a few years back. > > Using a phone with accelerometers they managed to identify different > > persons from their walk with about a 90% accuracy IIRC. So it seems to > > be completely feasible, but I doubt that it's trivial. > > > > also this research paper: > "Footstep identification from pressure signals using hidden markov models" > http://www.ee.oulu.fi/mvg/files/pdf/pdf_459.pdf You folks are freaking me out. Outside of tracking people, any other ideas for why you'd want this? -- H. Lally Singh Ph.D. Candidate, Computer Science Virginia Tech ___ Openmoko community mailing list community@lists.openmoko.org http://lists.openmoko.org/mailman/listinfo/community
Re: Accelerometer brainstorming
On Lun 31 mars 2008 10:59, Kalle Happonen a écrit : > Alexey Feldgendler wrote: >> >> Do you think it's possible to use traits of a person's walk for >> identification? Never heard about something like this. Interesting >> idea, if it turns out implementable. >> > I remember seeing some research results about this a few years back. > Using a phone with accelerometers they managed to identify different > persons from their walk with about a 90% accuracy IIRC. So it seems to > be completely feasible, but I doubt that it's trivial. > also this research paper: "Footstep identification from pressure signals using hidden markov models" http://www.ee.oulu.fi/mvg/files/pdf/pdf_459.pdf Gilles ___ Openmoko community mailing list community@lists.openmoko.org http://lists.openmoko.org/mailman/listinfo/community
Re: Accelerometer brainstorming
Alexey Feldgendler wrote: Do you think it's possible to use traits of a person's walk for identification? Never heard about something like this. Interesting idea, if it turns out implementable. I remember seeing some research results about this a few years back. Using a phone with accelerometers they managed to identify different persons from their walk with about a 90% accuracy IIRC. So it seems to be completely feasible, but I doubt that it's trivial. Kalle ___ Openmoko community mailing list community@lists.openmoko.org http://lists.openmoko.org/mailman/listinfo/community
Re: Accelerometer brainstorming
Another idea would be to use the accelerometers as a method of text entry in combination with dasher [1] - similar to this [2]. Could turn out to be a lot easier than using a small on-screen keyboard... [1] http://www.inference.phy.cam.ac.uk/dasher/ [2] http://www.inference.phy.cam.ac.uk/dasher/development/Tilt.html ___ Openmoko community mailing list community@lists.openmoko.org http://lists.openmoko.org/mailman/listinfo/community
RE: Accelerometer brainstorming
Ya, I started down the API path with Michael Shiloh a while back, but we got diverted by other things. Hi matt, Lets talk about your ideas, we need to highlight the new hardware we added. Let me know what you are thinking -Original Message- From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] On Behalf Of Crane, Matthew Sent: Friday, March 28, 2008 11:08 AM To: List for Openmoko community discussion Subject: RE: Accelerometer brainstorming Yea, there would need to be a decent test and experiment app, with GUI, record data feature, and visualization of data. That would be needed likely before the ideas we've mentioned could be worked out. Need to develop a solid API and event notification system. That sort of thing prob exists for GPS, but should it be new/separate or tagged onto GPS for acellerometer? Matt -Original Message- From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] On Behalf Of Ben Burdette Sent: Friday, March 28, 2008 1:40 PM To: List for Openmoko community discussion Subject: Re: Accelerometer brainstorming Alexey Feldgendler wrote: > Just like my recent post about using GPS to enhance the phone, I'll > try to explore the synergy opportunities for the accelerometers that > Neo has. Once again, some of this might have already been discussed. I'd like to see an accelerometer graphing application, so you could monitor and graph g force levels. Maybe throw in a zeroing button to set a baseline orientation, then show deviation from that. Maybe not extraordinarily useful, but it would be cool to look at. You could also record the data for later hacking or comparison, like what was my max g-force last time I went skydiving compared to this time? The graphing and etc would be good for studying the feasability of other accelerometer uses, ie is a tap on the case a recognizable event? Speaking of that, I'd like a tap on the case to be configurable as a mute function during ringing. So if you forgot to mute your phone at the symphony and it starts ringing somewhere inside your backpack or jacket, a well placed kick would (probably) mute it. The value here is not having to hunt for the phone and find a button on it. I guess this falls under the larger category of gesture recognition, but this alone would be a real win for me. Another use would be as a 0-60 or quarter mile acceleration timer. Here's a link to a company that makes these: http://www.gtechpro.com/?gclid=CLf7wPmpsJICFRI0awod1CWyQQ I don't know if the accelerometer in the freerunner is accurate enough, but this is a low priced product, I'd think their accelerometers are not a great deal better, if any. ___ Openmoko community mailing list community@lists.openmoko.org http://lists.openmoko.org/mailman/listinfo/community ___ Openmoko community mailing list community@lists.openmoko.org http://lists.openmoko.org/mailman/listinfo/community ___ Openmoko community mailing list community@lists.openmoko.org http://lists.openmoko.org/mailman/listinfo/community
RE: Accelerometer brainstorming
Yea, there would need to be a decent test and experiment app, with GUI, record data feature, and visualization of data. That would be needed likely before the ideas we've mentioned could be worked out. Need to develop a solid API and event notification system. That sort of thing prob exists for GPS, but should it be new/separate or tagged onto GPS for acellerometer? Matt -Original Message- From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] On Behalf Of Ben Burdette Sent: Friday, March 28, 2008 1:40 PM To: List for Openmoko community discussion Subject: Re: Accelerometer brainstorming Alexey Feldgendler wrote: > Just like my recent post about using GPS to enhance the phone, I'll > try to explore the synergy opportunities for the accelerometers that > Neo has. Once again, some of this might have already been discussed. I'd like to see an accelerometer graphing application, so you could monitor and graph g force levels. Maybe throw in a zeroing button to set a baseline orientation, then show deviation from that. Maybe not extraordinarily useful, but it would be cool to look at. You could also record the data for later hacking or comparison, like what was my max g-force last time I went skydiving compared to this time? The graphing and etc would be good for studying the feasability of other accelerometer uses, ie is a tap on the case a recognizable event? Speaking of that, I'd like a tap on the case to be configurable as a mute function during ringing. So if you forgot to mute your phone at the symphony and it starts ringing somewhere inside your backpack or jacket, a well placed kick would (probably) mute it. The value here is not having to hunt for the phone and find a button on it. I guess this falls under the larger category of gesture recognition, but this alone would be a real win for me. Another use would be as a 0-60 or quarter mile acceleration timer. Here's a link to a company that makes these: http://www.gtechpro.com/?gclid=CLf7wPmpsJICFRI0awod1CWyQQ I don't know if the accelerometer in the freerunner is accurate enough, but this is a low priced product, I'd think their accelerometers are not a great deal better, if any. ___ Openmoko community mailing list community@lists.openmoko.org http://lists.openmoko.org/mailman/listinfo/community ___ Openmoko community mailing list community@lists.openmoko.org http://lists.openmoko.org/mailman/listinfo/community
Re: Accelerometer brainstorming
Alexey Feldgendler wrote: Just like my recent post about using GPS to enhance the phone, I'll try to explore the synergy opportunities for the accelerometers that Neo has. Once again, some of this might have already been discussed. I'd like to see an accelerometer graphing application, so you could monitor and graph g force levels. Maybe throw in a zeroing button to set a baseline orientation, then show deviation from that. Maybe not extraordinarily useful, but it would be cool to look at. You could also record the data for later hacking or comparison, like what was my max g-force last time I went skydiving compared to this time? The graphing and etc would be good for studying the feasability of other accelerometer uses, ie is a tap on the case a recognizable event? Speaking of that, I'd like a tap on the case to be configurable as a mute function during ringing. So if you forgot to mute your phone at the symphony and it starts ringing somewhere inside your backpack or jacket, a well placed kick would (probably) mute it. The value here is not having to hunt for the phone and find a button on it. I guess this falls under the larger category of gesture recognition, but this alone would be a real win for me. Another use would be as a 0-60 or quarter mile acceleration timer. Here's a link to a company that makes these: http://www.gtechpro.com/?gclid=CLf7wPmpsJICFRI0awod1CWyQQ I don't know if the accelerometer in the freerunner is accurate enough, but this is a low priced product, I'd think their accelerometers are not a great deal better, if any. ___ Openmoko community mailing list community@lists.openmoko.org http://lists.openmoko.org/mailman/listinfo/community
RE: Accelerometer brainstorming
I don't think that's reaching at all! The blogger running http://gadgets.boingboing.net just blogged this week about how he lost his iPhone - it fell out of his pocket in a taxicab and he didn't realize until after. Maybe if it happens 6 months hence we can berate him for not avoiding the whole problem by having bought a FreeRunner instead. Bobby > * Detect when phone has dropped out of ones pocket, short fall by aprupt > stop, beep three times loudly. (that's kind of reaching..) > > Matt ___ Openmoko community mailing list community@lists.openmoko.org http://lists.openmoko.org/mailman/listinfo/community
Re: Accelerometer brainstorming
On Fri, Mar 28, 2008 at 06:18:51PM +0100, Alexey Feldgendler wrote: > On Fri, 28 Mar 2008 17:59:38 +0100, Crane, Matthew <[EMAIL PROTECTED]> > wrote: > >* Use acell data to charecterize person carrying the phone (as many > >couples out there will share phones, or give to children) and tie into > >user profile. > > Do you think it's possible to use traits of a person's walk for > identification? Never heard about something like this. Interesting idea, > if it turns out implementable. Gait analysis (from video footage) can be used to identify people[1]. Whether it's possible to do so using accelerometer data is an interesting question. Hugo. [1] http://www.gait.ecs.soton.ac.uk/ -- === Hugo Mills: [EMAIL PROTECTED] carfax.org.uk | darksatanic.net | lug.org.uk === PGP key: 515C238D from wwwkeys.eu.pgp.net or http://www.carfax.org.uk --- All mushrooms are edible, but some are only edible once. --- signature.asc Description: Digital signature ___ Openmoko community mailing list community@lists.openmoko.org http://lists.openmoko.org/mailman/listinfo/community
Re: Accelerometer brainstorming
On Fri, 28 Mar 2008 19:09:14 +0100, Peter Kraker <[EMAIL PROTECTED]> wrote: Unfortunately you need accurate speed to be able to get usable data from accelerometers for that. That's why some GPS modules have an interface for tachometer. If reading tachometer data from modern cars is possible with Nokia N770/N800, it is theoretically possible with Neo as well. http://openbossa.indt.org/carman/ -- Alexey Feldgendler <[EMAIL PROTECTED]> [ICQ: 115226275] http://feldgendler.livejournal.com ___ Openmoko community mailing list community@lists.openmoko.org http://lists.openmoko.org/mailman/listinfo/community
RE: Accelerometer brainstorming
On Fri, 2008-03-28 at 12:59 -0400, Crane, Matthew wrote: > * Detect when phone has dropped out of ones pocket, short fall by aprupt > stop, beep three times loudly. (that's kind of reaching..) Better even: start self check and if it fails play HAL singing "Daisy" while fading away (from Kubrick's Space Odyssey) :) ___ Openmoko community mailing list community@lists.openmoko.org http://lists.openmoko.org/mailman/listinfo/community
Re: Accelerometer brainstorming
On Fri, 28 Mar 2008 17:59:38 +0100, Crane, Matthew <[EMAIL PROTECTED]> wrote: * I mentioned this in another thread, but if the device: - goes from 60km+ to 0 in short order, e.g. high g stop - while traveling horizontally - over a road - an on-screen alert/countdown is not stoped Then it's likely a vehicular accident so auto-call/sms for help with some kind of countdown to disable. Only really possible to do that with a phone + GPS + acell. In some countries it's illegal to call emergency numbers automatically. * Use acell data to charecterize person carrying the phone (as many couples out there will share phones, or give to children) and tie into user profile. Do you think it's possible to use traits of a person's walk for identification? Never heard about something like this. Interesting idea, if it turns out implementable. -- Alexey Feldgendler <[EMAIL PROTECTED]> [ICQ: 115226275] http://feldgendler.livejournal.com ___ Openmoko community mailing list community@lists.openmoko.org http://lists.openmoko.org/mailman/listinfo/community
Re: Accelerometer brainstorming
Unfortunately you need accurate speed to be able to get usable data from accelerometers for that. That's why some GPS modules have an interface for tachometer. Joseph Reeves pravi: To join this post and your previous post, how about using the accelerometers to enhance the GPS? Not just for power saving, but say you drive through a tunnel; the GOS loses connectivity but the device still knows your location based on accelerometer data. There are commercially available in car navigation systems that already employ this technology. Joseph On 28/03/2008, Alexey Feldgendler <[EMAIL PROTECTED]> wrote: Just like my recent post about using GPS to enhance the phone, I'll try to explore the synergy opportunities for the accelerometers that Neo has. Once again, some of this might have already been discussed. * Intuitive mute: put the ringing phone on a flat horizontal surface (table) with its screen facing down to stop the ringing. Intuitive for the user and easy to detect with the accelerometers. Could also be used to hang up the current call. The gesture is very distinguishable and is hard to misinterpret because there aren't many situations when the user would normally put the phone on the table screen down. * Vibration control: when the phone is lying flat, either with its screen up or down, don't vibrate, so as to not produce the annoying loud noise. Ring instead, or do something else. Maybe make a weaker vibration, if the vibrator can do that. * Step counter: use the accelerometers to count steps when walking or running. Some people use dedicated devices for that. * GPS power saving: you can't move if you don't accelerate, so don't do power-expensive location detection until acceleration is detected. Once it's detected, start monitoring the location and keep doing it. Stop monitoring once the position stops changing for a while. -- Alexey Feldgendler <[EMAIL PROTECTED]> [ICQ: 115226275] http://feldgendler.livejournal.com ___ Openmoko community mailing list community@lists.openmoko.org http://lists.openmoko.org/mailman/listinfo/community ___ Openmoko community mailing list community@lists.openmoko.org http://lists.openmoko.org/mailman/listinfo/community ___ Openmoko community mailing list community@lists.openmoko.org http://lists.openmoko.org/mailman/listinfo/community
RE: Accelerometer brainstorming
* I mentioned this in another thread, but if the device: - goes from 60km+ to 0 in short order, e.g. high g stop - while traveling horizontally - over a road - an on-screen alert/countdown is not stoped Then it's likely a vehicular accident so auto-call/sms for help with some kind of countdown to disable. Only really possible to do that with a phone + GPS + acell. * Sense when the owner is in a car by keying on the vibration signature and acceleration, possibly with gps data, and automatically bring up the assigned travel/map application. * Use acell data to charecterize person carrying the phone (as many couples out there will share phones, or give to children) and tie into user profile. * Detect when phone has dropped out of ones pocket, short fall by aprupt stop, beep three times loudly. (that's kind of reaching..) Matt -Original Message- From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] On Behalf Of Alexey Feldgendler Sent: Friday, March 28, 2008 9:22 AM To: List for OpenMoko community discussion Subject: Accelerometer brainstorming Just like my recent post about using GPS to enhance the phone, I'll try to explore the synergy opportunities for the accelerometers that Neo has. Once again, some of this might have already been discussed. * Intuitive mute: put the ringing phone on a flat horizontal surface (table) with its screen facing down to stop the ringing. Intuitive for the user and easy to detect with the accelerometers. Could also be used to hang up the current call. The gesture is very distinguishable and is hard to misinterpret because there aren't many situations when the user would normally put the phone on the table screen down. * Vibration control: when the phone is lying flat, either with its screen up or down, don't vibrate, so as to not produce the annoying loud noise. Ring instead, or do something else. Maybe make a weaker vibration, if the vibrator can do that. * Step counter: use the accelerometers to count steps when walking or running. Some people use dedicated devices for that. * GPS power saving: you can't move if you don't accelerate, so don't do power-expensive location detection until acceleration is detected. Once it's detected, start monitoring the location and keep doing it. Stop monitoring once the position stops changing for a while. -- Alexey Feldgendler <[EMAIL PROTECTED]> [ICQ: 115226275] http://feldgendler.livejournal.com ___ Openmoko community mailing list community@lists.openmoko.org http://lists.openmoko.org/mailman/listinfo/community ___ Openmoko community mailing list community@lists.openmoko.org http://lists.openmoko.org/mailman/listinfo/community
Re: Accelerometer brainstorming
Alexey Feldgendler wrote: Just like my recent post about using GPS to enhance the phone, I'll try to explore the synergy opportunities for the accelerometers that Neo has. Once again, some of this might have already been discussed. * Intuitive mute: put the ringing phone on a flat horizontal surface (table) with its screen facing down to stop the ringing. Intuitive for the user and easy to detect with the accelerometers. Could also be used to hang up the current call. The gesture is very distinguishable and is hard to misinterpret because there aren't many situations when the user would normally put the phone on the table screen down. Very nice feature that would be great to see it in action. * Vibration control: when the phone is lying flat, either with its screen up or down, don't vibrate, so as to not produce the annoying loud noise. Ring instead, or do something else. Maybe make a weaker vibration, if the vibrator can do that. * Step counter: use the accelerometers to count steps when walking or running. Some people use dedicated devices for that. * GPS power saving: you can't move if you don't accelerate, so don't do power-expensive location detection until acceleration is detected. Once it's detected, start monitoring the location and keep doing it. Stop monitoring once the position stops changing for a while. * control menus and games with accelometer. Im new to this openmoko world but what do you think about 2d neverball clone on neo that you control with changing angle of phone position. How are accelometer signals captured by os? Any known application that usses accelometers? ___ Openmoko community mailing list community@lists.openmoko.org http://lists.openmoko.org/mailman/listinfo/community
Re: Accelerometer brainstorming
On Fri, 28 Mar 2008 14:31:20 +0100, Joseph Reeves <[EMAIL PROTECTED]> wrote: To join this post and your previous post, how about using the accelerometers to enhance the GPS? Not just for power saving, but say you drive through a tunnel; the GOS loses connectivity but the device still knows your location based on accelerometer data. There are commercially available in car navigation systems that already employ this technology. I haven't mentioned accelerometer-assisted dead reckoning because I remember it having already been discussed. The same goes for recognition of various gestures like shaking and tilting. -- Alexey Feldgendler <[EMAIL PROTECTED]> [ICQ: 115226275] http://feldgendler.livejournal.com ___ Openmoko community mailing list community@lists.openmoko.org http://lists.openmoko.org/mailman/listinfo/community
Re: Accelerometer brainstorming
To join this post and your previous post, how about using the accelerometers to enhance the GPS? Not just for power saving, but say you drive through a tunnel; the GOS loses connectivity but the device still knows your location based on accelerometer data. There are commercially available in car navigation systems that already employ this technology. Joseph On 28/03/2008, Alexey Feldgendler <[EMAIL PROTECTED]> wrote: > Just like my recent post about using GPS to enhance the phone, I'll try to > explore the synergy opportunities for the accelerometers that Neo has. > Once again, some of this might have already been discussed. > > * Intuitive mute: put the ringing phone on a flat horizontal surface > (table) with its screen facing down to stop the ringing. Intuitive for the > user and easy to detect with the accelerometers. Could also be used to > hang up the current call. The gesture is very distinguishable and is hard > to misinterpret because there aren't many situations when the user would > normally put the phone on the table screen down. > > * Vibration control: when the phone is lying flat, either with its screen > up or down, don't vibrate, so as to not produce the annoying loud noise. > Ring instead, or do something else. Maybe make a weaker vibration, if the > vibrator can do that. > > * Step counter: use the accelerometers to count steps when walking or > running. Some people use dedicated devices for that. > > * GPS power saving: you can't move if you don't accelerate, so don't do > power-expensive location detection until acceleration is detected. Once > it's detected, start monitoring the location and keep doing it. Stop > monitoring once the position stops changing for a while. > > > > -- > Alexey Feldgendler <[EMAIL PROTECTED]> > [ICQ: 115226275] http://feldgendler.livejournal.com > > ___ > Openmoko community mailing list > community@lists.openmoko.org > http://lists.openmoko.org/mailman/listinfo/community > ___ Openmoko community mailing list community@lists.openmoko.org http://lists.openmoko.org/mailman/listinfo/community