Post by thread:[EE]: Looking formethodtodifferentiategravityfroma

>> His key point was not the 3 axes, 2 axes would usually be OK as you >> say, but the fact that the vector sum of gravity will be 1g >> regardless >> of vehicle orientation. Anything above or below 1g MUST be from >> other >> sources. > I thought that at first, but you can have other forces and still > have a > 1g vector. As an extreme example, add a 2g vector in the opposite > direction. That example isn't included in my solution space. I said above or below 1g :-). I suppose that some acceleration sources could exceed 2g, but this is liable to be for extremely short periods. For example, a "judder bar" may well do interesting things. And it seems unlikely that any really major ones would be horizontal. Except maybe collisions, at which point the system has failed to do its job and may be retired while the paperwork is completed. RM

On Sat, Sep 03, 2005 at 02:29:09AM +1200, Russell McMahon wrote: > >>His key point was not the 3 axes, 2 axes would usually be OK as you > >>say, but the fact that the vector sum of gravity will be 1g > >>regardless > >>of vehicle orientation. Anything above or below 1g MUST be from > >>other > >>sources. > > >I thought that at first, but you can have other forces and still > >have a > >1g vector. As an extreme example, add a 2g vector in the opposite > >direction. > > That example isn't included in my solution space. > I said above or below 1g :-). True, when it's != 1g you know there's some acceleration, but when it's 1g there might still be some. > I suppose that some acceleration sources could exceed 2g, but this is > liable to be for extremely short periods. The 2g was just an extreme example. A smaller acceleration could also result in a total of 1g in a different direction, of course. Chris

Chris Emerson wrote: >>> I thought that at first, but you can have other forces and still >>> have a >>> 1g vector. As an extreme example, add a 2g vector in the opposite >>> direction. >> >> That example isn't included in my solution space. >> I said above or below 1g :-). > > True, when it's != 1g you know there's some acceleration, but when it's > 1g there might still be some. This is because you are treating a vector as a scalar. The 3 axis system works in theory if you can subtract the gravity accelleration vector. Unfortunately while its magnitude is easy (1g), it's direction is not so easy to know without essentially doing an inertial navigation solution. While this is an interesting theoretical discussion, I don't think something like this is practical for a car to determine braking. I've also got serious concerns about the desirability of blinking a brake light in the first place. ***************************************************************** Embed Inc, embedded system specialists in Littleton Massachusetts (978) 742-9014, http://www.embedinc.com

The inertial navigation thing was done with an ATMEL chip on the First robot and again with a PIC 16F877. By orientation of the 3 axis accelerometer you can make some assumptions. If you are accelerating straight down at 1 G the least of your problems is blinking the brake light. ;) Also the acceleration vector would change fairly quickly in a brake situation. You use more than just magnitude. You use the angle as well. not extremly difficult to solve nor over expensive. A lot easier than determining the position using integration of the acceleration as we did on the robot. Larry {Quote hidden}> > From: spam_OUTolin_piclistTakeThisOuTembedinc.com (Olin Lathrop) > Date: 2005/09/02 Fri AM 11:42:28 EDT > To: "Microcontroller discussion list - Public." <.....piclistKILLspam@spam@mit.edu> > Subject: Re: RE: [EE]: Looking formethodtodifferentiategravityfromacceleration > > Chris Emerson wrote: > >>> I thought that at first, but you can have other forces and still > >>> have a > >>> 1g vector. As an extreme example, add a 2g vector in the opposite > >>> direction. > >> > >> That example isn't included in my solution space. > >> I said above or below 1g :-). > > > > True, when it's != 1g you know there's some acceleration, but when it's > > 1g there might still be some. > > This is because you are treating a vector as a scalar. The 3 axis system > works in theory if you can subtract the gravity accelleration vector. > Unfortunately while its magnitude is easy (1g), it's direction is not so > easy to know without essentially doing an inertial navigation solution. > > While this is an interesting theoretical discussion, I don't think something > like this is practical for a car to determine braking. I've also got > serious concerns about the desirability of blinking a brake light in the > first place. > > > ***************************************************************** > Embed Inc, embedded system specialists in Littleton Massachusetts > (978) 742-9014, http://www.embedinc.com > --

Olin Lathrop wrote: > While this is an interesting theoretical discussion, I don't think > something like this is practical for a car to determine braking. I've > also got serious concerns about the desirability of blinking a brake > light in the first place. That's what it seems to me, too. IMO the purpose of the brake light is to indicate that the driver is braking -- which seems to be perfectly addressed by switching it on when the driver is stepping on the brake pedal :) OTOH, there are probably situations where stepping of the gas can yield a brake-like effect, like when going uphill. In such a situation, a deceleration-controlled brake light may make sense. This would point more in the direction of doppler-based or similar speed measurements, rather than accelerometer-based ones. Gerhard