Hi guys my final year project is to detect the firemen location in a firescene (in a building without gps)
I plan to use an accelerometer and gyroscope to estimate the location of firemen in a building and using a microcontroller like arduino to process the raw dara and further display it on the computer, so that the commander can know where the firemen is when anything happens and can command other firemen to rescue the firemen

Will my idea works? It sounds too good to be true, if it doesnt work , what other solution can i use?

 

Early versions based on mechanical analog systems went screaming into London on top of V2 rockets during the later stages of World War II.

The idea is very practicable and perhaps ultimately marketable.

I wouldn't be so concerned with the inertial tracker but with the radio (?) communications needed to tell the home base where someone is.

 

Actually, the V2 rockets didn't know where they were. They only had a predictable distance. The Brits successfully diverted some of the rockets by sending false information about where the rockets landed. The Germans adjusted the firing direction according to the false information and missed London with some of their rockets.

Still, inertial guidance has been accomplished since then and can be used to locate an object.
(That doesn't mean I know how to do it.)

 

It is called "dead reckoning" navigation. If you know your starting location, speed, direction and time transpired you can determine your new location.

Obviously, one can think immediately of four ways why the results would be inaccurate.

 

Hi guys my final year project is to detect the firemen location in a firescene (in a building without gps)
I plan to use an accelerometer and gyroscope to estimate the location of firemen in a building and using a microcontroller like arduino to process the raw dara and further display it on the computer, so that the commander can know where the firemen is when anything happens and can command other firemen to rescue the firemen

Will my idea works? It sounds too good to be true, if it doesnt work , what other solution can i use?

How about a wearable device using RFID?
I believe RFID is capable of real-time 3D positioning...

eT

 

Isn't the Nike Plus system essentially based on this type of sensing? (accelerometer)

 

The issue is accelerometer noise and the second order integration that is required to find distance from acceleration. Some very clever people ended up spending an awful lot of money launching a global network of satellites because they couldn't improve this problem.

 

Isn't the Nike Plus system essentially based on this type of sensing? (accelerometer)

That device uses GPS.

The OP stated "in a building without GPS". Or maybe I'm misunderstanding what the OP really means(?).

eT

 

That device uses GPS.

Definitely not the early versions. It was accelerometers. Sorry if I linked to the wrong device. It's very hard to sort out what technologies are employed in these things. My daughter's FitBit is another example of a smart-pedometer, ie. accelerometer.

 

so according to what the previous post suggested, in my case RFID is a better solution ?

 

can i use a imu module to track firemen location ?

 

Probably not.

Experts try it with robots, that move on wheels at very steady speeds and known accelerations, and on smooth surfaces etc in controlled known environments. Dead reckoning still gets really messed up after a few moves and turns.

Your ability to get something workable on a walking/dodging/climbing fireman in a complex unknown environment will be practically zero.

 

There are some fairly stable IMU's available such as this but you would need to determine what accuracy and stability you would need for your location accuracy requirements. Then you need a microprocessor to calculate the person's location in 3 dimensions from the IMU data, and then some radio communication unit to broadcast the location to a receiver outside the building.

And how would you translate that 3d location in space to a particular location in the building? You would need a 3d layout of the building for that.

 

I would agree that this isn't feasible, in the context of a final year project.

However, perhaps you could achieve a proof-of-concept result with 2D navigation in a small room, over a reasonably short period of time, if you calibrate your sensors very carefully.

The latest relevant and applicable research that I know of is here: http://www.x-io.co.uk/category/open-source/

 

I doubt that RFID has the range you need but I would have thought that you could try a combination of an IMU together with a beacon system - i.e. the fire-fighter carries a radio beacon and two or three fire trucks have directional antennas that detect the individual beacons. The IMU can be used to fill in the fine details between beacon measurements.

 

Can you further explain it? It seems interesting

 

I suppose what I had in mind is that every firefighter has a radio transmitter that sends out a code (a bit like RFID but powered so it has higher output power). Assuming the firefighters are in a building, two fire trucks are parked at say either corner of the building, directional antenna on the trucks will give an angle to the firefighter, knowing the distance between the trucks then simple trigonometry will give the distance to the firefighter.

However, knowing the firefighter's distance from base is not the whole picture, if you are going to go in and pull out a trapped firefighter you need to know how to get to her, this is where the IMU comes in. It will tell you roughly the route they took, e.g. turned left, east along corridor, up stairs, etc.

 

...

However, perhaps you could achieve a proof-of-concept result with 2D navigation in a small room, over a reasonably short period of time, if you calibrate your sensors very carefully.
...

Perhaps, with the right level of skill.

But from inertial navigation alone it's worse than just sensor calibration. Every turn and course change introduces more error because of the sensor resolution, data rounding etc. The errors multiply exponentially as more moves are made.

It gets closer to workable if you can add in a non-inertial (absolute) data source, like a magnetic compass etc. Even though it has low resolution to read absolute angle it can be used to stop the inertial calc errors growing out of bounds. In robotics they use other absolute data as well, like knowlege of the walls and what angle the walls are in relation to north etc.

 

the only reason for not using gps is that the signals might not penetrate the building. if the gps cant penetrate the building, how can an r beacon? how would the inertial system send its data? why dosnt somebody develop a system using gps methods that is used with local transmitters instead of satelites?

 

Higher frequencies do not penetrate buildings as well as lower frequencies so by using a lower frequency the beacon can penetrate the building. After all, fire/police/etc. walkie-talkies work in buildings.

GPS is very expensive technology, it needs very accurate clocks, hence why no one has developed a local version.