programming deadreckoning gps software?

**Dominik** · 09-28-2004, 06:40 PM

Dunno about the states, but we have an electronics store called Jaycar here, and they sell an overspeed warning kit which uses a small magnet attached to your axel or something like that to get the speed reading. It was only like $20 AUD i think; perhaps thats an option?

Also, most late model cars use digital speedo signals taken from the gearbox. All you need is something to count the pulses and derive a speed from that. You'll be able to find the wire with a Haynes manual or something like that...

**Ricky327** · 09-28-2004, 08:00 PM

All you need is something to count the pulses and derive a speed from that.

But getting the speed directly from it may not be good enough since I dont think it can register a car speed when in reverse...going forward/reverse in a carpark lets say.

Ideally 2 sensors should be used to detect the speed and direction of the road wheel rotation. The rotation depend on which sensor is getting triggered first...and of course the speed is proportional to the frequency of the pulse. As been said magnetic sensors are ideal to use for cars.

jpelzer,

Did you get to try out the phigets? just out of interest does it register a tilt? or?

**Ricky327** · 09-28-2004, 08:03 PM

Without any phidgets, the simplest solution would be to extrapolate the coordinates by using the last known direction/speed. Wouldnt be accurate for more than 1-2 seconds, or if you are turning etc, but better than nothing...

I though some GPS receiver already does this?

**Dominik** · 09-28-2004, 10:45 PM

Not sure... To tell the truth, i dont have an operational GPS receiver (still building it using a Trimble GPS board), but i doubt mine will have that.

Since i'm building it myself i am hoping i can implement this through hardware (i will need help though!!!)

**jpelzer** · 09-29-2004, 12:44 AM

Originally Posted by Ricky327

Did you get to try out the phigets? just out of interest does it register a tilt? or?

I've got the interface to the phidget going just fine. Definitely registers tilt. Lay it flat, it measures 0g in both axis. On its side, 1g in one axis, 0g in the other... etc.

Since the sensor doesn't work as a had hoped going back to the very beginning of the thread, and the tilt of the car is definitely going to affect the reading, I'm only planning on using the one axis for approximate steering data, and then using the OBDII speed data to tie it together.

It will definitely be a rough approximation. But it should work better than nothing through the instances that I am designing for: Tunnels, lower bridge levels (GWB) and highway exits under overpasses.

It's not going to work on a two-level freeway (at least not well). Or in a parking garage (not for long, anyways). A far as two-level freeways go, I don't have any of them around here to test on. Closest thing I have is the lower-deck of the George Washington Bridge.

After playing with the linear accelerometer, I have serious misgivings about how well it will work for this application. So I'm hoping others have luck with either the electronic compass, or we find an inexpensive supplier for a USB-connected radial accelerometer.

**vision** · 09-29-2004, 04:36 PM

Guys, do not recreate wills. Gyro + Speed sensor is optimal for price/precision. Speed sensor present on every car, gyro price, like $40-50. Microchip for calculate position and regenerate NMEA stream $1.5. With factory printed board and wires/accesory and connectors price for device max $150.

With regards.

**Ricky327** · 09-29-2004, 04:40 PM

Guys, do not recreate wills. Gyro + Speed sensor is optimal for price/precision.

I agree but where can we buy for that price? please post a link.

If theres already a PIC project then it will save so much hassle. For 150USD cant really go wrong.

**jpelzer** · 09-29-2004, 05:21 PM

Yeah, the whole thread has basically agreed that a radial accelerometer is the right way to go... Just to clarify, when we say "radial accelerometer" it is the same as "gyro". I think we're all referring to a piezo-electric rate accelerometer.

And yes, the chips are about $40-50 if you can find them. The only place I found for these chips is selling a model that has been discontinued by the manufacturer.

**rando** · 09-29-2004, 05:59 PM

Wills = wheels?

**rando** · 10-17-2004, 03:07 AM

Originally Posted by rando

I have that one and some code I wrote to read it over the parallel port. I haven't tried it in my car yet though I'm bit concerned that it might not be too accurate because of all the metal and other magnetic interference in the car. Walk up to your car with a normal compass and check our what happens. Maybe that's why dead reckoning is usually done with gyros?

I finally got around to trying the CMPS03 compass in my truck and, as expected, it was useless. I tried several positions with pretty much the same results. Rotating (or moving) the compass in the truck would change the reading. Moving the truck and compass together though produced minimal change on the reading -- in other words the compass reads the magnetic field produced by the truck instead of the earth.

I know a gyro based solution will work but I'm still curious about one thing. Does anyone know how the compass mirrors work internally? They are somehow able to determine direction without being rendered useless due to their proximity to a vehicle. What gives?