spenczar

I have a 2-axis accelerometer circuit that will be sending values to max/msp, and I'd like to be able to determine location information over time. The last time I used calculus was about 15 years ago, so I'm a little rusty here, but I think I'd need to integrate to get velocity and then displacement.

Is this something that's possible in Max? Any pointers would be greatly appreciated!!

I have a 2-axis accelerometer circuit that will be sending values to max/msp, and I'd like to be able to determine location information over time.  The last time I used calculus was about 15 years ago, so I'm  a little rusty here, but I think I'd need to integrate to get velocity and then displacement.  

Is this something that's possible in Max?  Any pointers would be greatly appreciated!!

integrate-to-get-displacement-from-acceleration-over-time

I tried this before, but didn't get accurate results and stopped.

The idea in integration is to sum the difference between consecutive 

values of your data stream multiplied by the duration of the interval 

between these values (like computing a surface):

-> sum [ (value[timeX]-value[timeX-1])*(timeX-(timeX-1)) ]

should give you the speed. And if you apply the same process on speed, 

You probably won't get your data on a regular time interval, so 

The problem is that the datas, and especially the time intervals, are 

not extremely accurate, so after a while, your speed won't be 0 

although the accelerometer doesn't move anymore because the sum of 

acceleration and deceleration won't _perfectly_ match. And so the 

As an accelaration of 0 doesn't mean that there is no movement (I was 

surprised to notice that the violin player I worked with moved her bow 

with a perfectly stable speed, i.e. no acceleration!), it is not easy 

to add some "reset" function to the speed and position; it depends on 

the physical properties of your moving object (in the case of a bow, 

you know the range of the movement, and could eventually manually reset 

the position when an extremity of the bow is reached).

Another issue is that your accelerometer will probably return 0->1 G 

values if it's only rotated, without changing its position within the 

Centre de Recherches et de Formation Musicales de Wallonie asbl

The idea in integration is to sum the difference between consecutive 
values of your data stream multiplied by the duration of the interval 
between these values (like computing a surface):
-> sum [ (value[timeX]-value[timeX-1])*(timeX-(timeX-1)) ]

should give you the speed. And if you apply the same process on speed, 
you'll get the position.

You probably won't get your data on a regular time interval, so 
timeX-(timeX-1) will vary over time.

The problem is that the datas, and especially the time intervals, are 
not extremely accurate, so after a while, your speed won't be 0 
although the accelerometer doesn't move anymore because the sum of 
acceleration and deceleration won't _perfectly_ match. And so the 
positions will move too.

As an accelaration of 0 doesn't mean that there is no movement (I was 
surprised to notice that the violin player I worked with moved her bow 
with a perfectly stable speed, i.e. no acceleration!), it is not easy 
to add some "reset" function to the speed and position; it depends on 
the physical properties of your moving object (in the case of a bow, 
you know the range of the movement, and could eventually manually reset 
the position when an extremity of the bow is reached).

Another issue is that your accelerometer will probably return 0->1 G 
values if it's only rotated, without changing its position within the 
world.

_____________________________
Patrick Delges

Centre de Recherches et de Formation Musicales de Wallonie asbl
http://users.skynet.be/crfmw/max


On 24 avr. 06, at 13:00, Patrick Delges wrote:

Ooooppppsssss, I'm afraid I wasn't very concentrate when I posted this 

message and made a strange mix between integration and derivation...

You don't want to sum the differences between values, but the values 

themselves, times the delta-time. I found a small test patch I made 

that may give you some ideas... but there is still a lot to do!

Ooooppppsssss, I'm afraid I wasn't very concentrate when I posted this 
message and made a strange mix between integration and derivation...
You don't want to sum the differences between values, but the values 
themselves, times the delta-time. I found a small test patch I made 
that may give you some ideas... but there is still a lot to do!

Thanks for your reply! I will be testing this week, and I'll let you know how it works out.

Thanks for your reply!  I will be testing this week, and I'll let you know how it works out.

FYI...going from acceleration to position is going to be tricky as

you're going to have a second order integrator. Errors will

accumulate very fast without some sort of error correction term in

FYI...going from acceleration to position is going to be tricky as
you're going to have a second order integrator.  Errors will
accumulate very fast without some sort of error correction term in
your calculations.

> I have a 2-axis accelerometer circuit that will be sending values to

> max/msp, and I'd like to be able to determine location information

> over time. The last time I used calculus was about 15 years ago, so

> I'm a little rusty here, but I think I'd need to integrate to get

Mathematically its not a problem as Patrick showed, but as your

accelerometers are analog you will get errors which add up fast. You

won't be able to track absolute positions with that very precisely,

unless you combine it with a way to correct it, for example with a

camera, or with single events which could only occur at a defined point.

All depends on what you want to do with it.

spencer kiser wrote:
> I have a 2-axis accelerometer circuit that will be sending values to
> max/msp, and I'd like to be able to determine location information
> over time.  The last time I used calculus was about 15 years ago, so
> I'm  a little rusty here, but I think I'd need to integrate to get
> velocity and then displacement.

Mathematically its not a problem as Patrick showed, but as your
accelerometers are analog you will get errors which add up fast. You
won't be able to track absolute positions with that very precisely,
unless you combine it with a way to correct it, for example with a
camera, or with single events which could only occur at a defined point.

  [][]  [][][]  [][]  [][][]
[][][][][][][][][][][][][][][]

         Stefan Tiedje
         Klanggestalter
     Electronic Composition
               &
         Improvisation

            /~~~~~
     \   /|() ()|
     ))))) )|  |  |( \
     ///     _/)/ )))))
             ___/   ///

-------------------------x----
--_____-----------|-----------
--(_|_ ----|-----|-----()----
-- _|_)----|-----()-----------
----------()------------x-----

14, Av. Pr. Franklin Roosevelt,
94320 Thiais, France
Phone at CCMIX +33-1-57 42 91 09


integrate to get displacement from acceleration over time