Need help with sensors

Discussion in 'Tech Discussion' started by Mr Pancakes, Mar 9, 2018.

Tags:
  1. Recouer

    Recouer Well-Known Member

    Joined:
    Oct 31, 2016
    Messages:
    614
    Likes Received:
    524
    Reading List:
    Link
    ... at least when you're doing "basic physics" please don't make mathematic errors...
    when you fall from 1 meter, you get the relation 10*t^(2)=1 thus t=10^(-1/2)~0.3
    then 10*0.3=3m/s or 1km/hour or 0.5mph
    not 10mph!!
    (for the record: 10 mph would be a 33feet tall fall)


    alright, let's assume our body fall from 5 feet, or 1.5 meter and let's say that due to the softness or the body, it takes 0.1second for the body to come at a stop, wich is a very unprobable case as usually, the acceleration felt would be higher, since it is not necessarily a normal but we take into account the "worst case scenario" for the detector to be triggered.
    and since all the accumulated velocity must be opposed (i don't know how to say this but let's say that 10*t(fall)=a*t(impact) is true where a is the acceleration felt during the fall)
    now we just find t(fall) and we easily find the minimum acceleration felt.
    so we fell at an acceleration rate of 10m/s^2 for 1.5 meter, we thus get the relation t(fall) = (1.5/10)^(1/2)~0.47 (well i didn't want to use my calculator but...)
    and we had t(impact) at 0.1s thus we get a=47m/s^2 (and if you want a violent impact, multiply that by 5)
    of course, that's not necessarily very high, but if you equip your sensor with a button that will stop false alarm, you could get a rather reliable device.

    and when you say that it does not work 1 out of 20 times for me it rather mean that it will save your ass 19times out of 20 and i think that the 1 out of 20 is a risk i'd be willing to take.

    well too bad but a sensor is not used to reduce the "risk" of a fall but its "fatality" as when you're old and alone you won't necessarily be rescued if you have a fall
     
    Mr Pancakes likes this.
  2. Truerror

    Truerror Well-Known Member

    Joined:
    Jan 10, 2016
    Messages:
    432
    Likes Received:
    186
    Reading List:
    Link
    This. For much longer distances, you'd want a barometer.

    Btw, "close distance" here doesn't necessarily mean a few centimeters. It'd also work for a distance of several meters. Of course, if you're talking a few hundred meters, or even kilometers, then a barometer would be more suitable.
     
  3. gangbuntu

    gangbuntu Well-Known Member

    Joined:
    Jul 9, 2016
    Messages:
    120
    Likes Received:
    68
    Reading List:
    Link
    +1

    +1

    food for thoughts:
    - you'll need a sensor with gyro and accelero like GY521 - MPU-605 around us$4.52
    - accelero would give acceleration at 3 axis, with (almost) constant 9.8m/s downward (gravity)
    --- so whatever downward acceleration calculated by your system should factor the gravity out
    - if you put your sensor at glove/wristband, the "down" axis would keep on changing (between the sensor's 3 axis).
    --- this is why you need gyro to translate accel 3 axis into earth's front-left-down axis
    - you need to know that this kind of el-cheapo sensor drifts a lot
    --- this is why htc vive plants lots of sensors and fuse the reading.

    how to differentiate between sitting down and free falling?
    or even wild bat/golf swinging etc?
    2 things that come to mind:
    - combine with gyro + other sensors to extract the down acceleration
    - after the initial "maybe_falling" acceleration threshold is breach, keep on storing the down acc value.
    - analyze the data after predefined period &/ the acceleration falls below the threshold
    --- if ample deceleration were recorded it would most likely due to activity
    --- --- might yield false negative if the wearer falls on top of shock absorbing material (e.g: fluffy bed)
    --- if the acceleration stopped abruptly it would most likely the wearer has reach his/her destination (after falling)
    -- --- might yield false positive if the wearer performs jerky motions
     
    Mr Pancakes likes this.
  4. Mr Pancakes

    Mr Pancakes Well-Known Pancake

    Joined:
    May 9, 2016
    Messages:
    617
    Likes Received:
    308
    Reading List:
    Link
    Thanks, I decided not to build this, it's too complicated for a school project so I will build something else.
     
  5. Recouer

    Recouer Well-Known Member

    Joined:
    Oct 31, 2016
    Messages:
    614
    Likes Received:
    524
    Reading List:
    Link
    :facepalm: (if there was a head banging blob i'd most likly be using it)
    it's not that hard, if it's just a project that won't have any real application.
    you just need to consider 3 things:
    #where to wear the sensor for it to be the least affected by your movement (here the nearest possible to the center of gravity of a human body)
    #the position of the wearer when afflicted by a shock(that will determine a shock threshold)
    =>it would thus be better for it not to flip over when sitting
    #the shock threshold that has to depend on the position of the wearer, for that, you'll need the weight of the wearer as well as medical information to know at which point the force applied can prove dangerous.
     
    Mr Pancakes likes this.