Accelerometer measurements & Rowperfect
magnus....@googlemail.com
“On-Board-Shell Measurements of Acceleration” at
http://neutrino.phys.washington.edu/~wilkes/post/temp/phys208/shell.acceleration.html
. Several times he draws attention to a common feature of many or most
rowing-stroke shell acceleration plots: namely a “mystery dip in
acceleration at 70-90 degrees (relative to the forward direction)”. He
speculates at the end that it might be symptomatic of the blade
stalling and suggests experimenting with different rigs etc to try to
pin it down.
Meantime, I got thinking on reading the paper that I could (get a
friend to) build one of those and do some experimentation for myself.
I did, and I now have data for several scullers (and a 2x) doing
various different types of rowing.
My “toy” is a 1000Hz datalogger into which can be plugged two
thumbnail-sized +/-2g inertial accelerometers. Generally I put one on
the shell and one on the seat. One benefit of this arrangement is
that, by subtracting the seat’s acceleration from the shell’s and
integrating for relative velocity, it is possible to determine the
exact moment at which seat velocity is zero: i.e. frontstops. The
datalogger has enough memory for 4 minutes of data from the two
inputs.
Sure enough, the “mystery dip” seems to be an integral feature of the
way in which a shell gets moved by a sculling stroke. To a greater or
lesser extent, every single recorded stroke from every boat in my data
exhibits it. Until yesterday, my thoughts were that perhaps it was
evidence of the blade stalling and transitioning between “propeller to
pry-bar” as Zeke put it recently; or perhaps it was some resonance or
“bounce-back” of the oar-shaft after the initial catch.
I went to see Carl Douglas a couple of days ago at his factory to get
some seats repaired but also to show him the toy and some of the data
collected so far. He pointed to a Rowperfect sitting nearby and
suggested sticking an accelerometer on that. So I did, and do you know
what? That “mystery dip” shows up on the floating head of the
Rowperfect too. I’ve posted a couple of images at www.slidingseat.net
This, and the fact that the Rowperfect signal is in all other respects
very similar to the sculling signal, suggests a couple of things to
me:
1 The Rowperfect is a not-bad simulator for rowing/sculling
2a The post-catch acceleration hump/dip may be bio-mechanical rather
than an artefact of rigging/materials
2b The dip may still be an artefact of rigging/materials, but
instilled into my muscle-memory by 27 years of rowing such that I
repeat it even when I’m on a machine.(i.e I’ve recorded the signal to
my body, and I play it back on the simulator)
Comments appreciated
Cheers, Magnus
Donal
Sounds interesting. Perhaps repeat the experiments on the rowperfect
with non-rowers to get over the muscle memory idea.
Does the dip occur when doing different phases...1/4 slide , half
slide, straight arm rowing?
Regards
Donal
magnus....@googlemail.com
yes getting an experienced non-rower erger would be useful. I do have
some backstops paddling to 1/4 to 1/2 slide data and the feature
appears there too, but the noise/signal ratio is higher. I plan on
doing it and other outings again with a cleaner installation.
zeke_hoskin
> Many of you may have seen a short paper by Ken Young in 1991 entitled
It's tempting to suggest the obvious. A graph of shell/rowing head
acceleration is meaningless without a corresponding graph of rower and/
or center-of-mass acceleration. The mystery dip in shell acceleration
would then correspond to a point in the stroke where leg drive and
back unfolding overlap, so body mass accelerates forward wrt the shell
and shell mass decelerates. Alternatively, it's the point where leg
drive ends and back angle is still awkward from a force-application
point of view. Lining up the acceleration curves with the rigger force
curves would quickly show which after-the-fact guess/explanation
applies.
Of course, if that were all there was to it, rowing out-of-phase would
be faster than rowing in sync. . .//Zeke Hoskin
Carl Douglas
> On Jun 5, 7:05 am, magnus.burba...@googlemail.com wrote:
>> Many of you may have seen a short paper by Ken Young in 1991 entitled
>> . Several times he draws attention to a common feature of many or most
>> acceleration at 70-90 degrees (relative to the forward direction)�. He
>> speculates at the end that it might be symptomatic of the blade
>> stalling and suggests experimenting with different rigs etc to try to
>> pin it down.
>>
>> Meantime, I got thinking on reading the paper that I could (get a
>> friend to) build one of those and do some experimentation for myself.
>> I did, and I now have data for several scullers (and a 2x) doing
>> various different types of rowing.
>>
>> thumbnail-sized +/-2g inertial accelerometers. Generally I put one on
>> the shell and one on the seat. One benefit of this arrangement is
>> integrating for relative velocity, it is possible to determine the
>> exact moment at which seat velocity is zero: i.e. frontstops. The
>> datalogger has enough memory for 4 minutes of data from the two
>> inputs.
>>
>> way in which a shell gets moved by a sculling stroke. To a greater or
>> lesser extent, every single recorded stroke from every boat in my data
>> exhibits it. Until yesterday, my thoughts were that perhaps it was
>> pry-bar� as Zeke put it recently; or perhaps it was some resonance or
>> �bounce-back� of the oar-shaft after the initial catch.
>>
>> I went to see Carl Douglas a couple of days ago at his factory to get
>> some seats repaired but also to show him the toy and some of the data
>> collected so far. He pointed to a Rowperfect sitting nearby and
>> suggested sticking an accelerometer on that. So I did, and do you know
>> Rowperfect too. I�ve posted a couple of images atwww.slidingseat.net
>>
>> This, and the fact that the Rowperfect signal is in all other respects
>> very similar to the sculling signal, suggests a couple of things to
>> me:
>>
>> 1 The Rowperfect is a not-bad simulator for rowing/sculling
>>
>> 2a The post-catch acceleration hump/dip may be bio-mechanical rather
>> than an artefact of rigging/materials
>>
>> 2b The dip may still be an artefact of rigging/materials, but
>> instilled into my muscle-memory by 27 years of rowing such that I
>> my body, and I play it back on the simulator)
>>
>> Comments appreciated
>>
>> Cheers, Magnus
>
> It's tempting to suggest the obvious. A graph of shell/rowing head
> acceleration is meaningless without a corresponding graph of rower and/
> or center-of-mass acceleration. The mystery dip in shell acceleration
> would then correspond to a point in the stroke where leg drive and
> back unfolding overlap, so body mass accelerates forward wrt the shell
> and shell mass decelerates. Alternatively, it's the point where leg
> drive ends and back angle is still awkward from a force-application
> point of view. Lining up the acceleration curves with the rigger force
> curves would quickly show which after-the-fact guess/explanation
> applies.
>
> Of course, if that were all there was to it, rowing out-of-phase would
> be faster than rowing in sync. . .//Zeke Hoskin
Having been present for this erg test, I can confirm that it was a
simple test, quickly set up. Magnus was in street clothing and not
warmed up, but the data and the ease of its generation were most impressive.
In what sense are the data produced so readily dismissed as
"meaningless". Sure, it'd be good to have more data, but there seems to
be plenty of value already - if one is prepared to do a bit of head
scratching.
Perhaps one way to take this forward under laboratory conditions, if the
apparent close similarity between the results obtained from Rowperfect
erg & boat is sustained after further experiments, would be to add
side-view video, with markers on key body locations, to RP simulations.
As for rigger loads, RP's monitor gives a continuous reading of stroke
force which, if coordinated with all the other readings, may give plenty
to play with.
An obvious benefit of the kind of data that Magnus is able to generate
is that a coach or rower might be able now to explore ways to smooth out
those lumps & dips, which may reveal their causes & to what extent they
are either inevitable or the result of poor existing technique.
Cheers -
Carl
--
Carl Douglas Racing Shells -
Fine Small-Boats/AeRoWing Low-drag Riggers/Advanced Accessories
Write: Harris Boatyard, Laleham Reach, Chertsey KT16 8RP, UK
Find: http://tinyurl.com/2tqujf
Email: ca...@carldouglas.co.uk Tel: +44(0)1932-570946 Fax: -563682
URLs: www.carldouglas.co.uk (boats) & www.aerowing.co.uk (riggers)
zeke_hoskin
> zeke_hoskin wrote:
> > On Jun 5, 7:05 am, magnus.burba...@googlemail.com wrote:
> >> Many of you may have seen a short paper by Ken Young in 1991 entitled
> >> . Several times he draws attention to a common feature of many or most
> >> acceleration at 70-90 degrees (relative to the forward direction)”. He
> >> speculates at the end that it might be symptomatic of the blade
> >> stalling and suggests experimenting with different rigs etc to try to
> >> pin it down.
>
> >> Meantime, I got thinking on reading the paper that I could (get a
> >> friend to) build one of those and do some experimentation for myself.
> >> I did, and I now have data for several scullers (and a 2x) doing
> >> various different types of rowing.
>
> >> thumbnail-sized +/-2g inertial accelerometers. Generally I put one on
> >> the shell and one on the seat. One benefit of this arrangement is
> >> integrating for relative velocity, it is possible to determine the
> >> exact moment at which seat velocity is zero: i.e. frontstops. The
> >> datalogger has enough memory for 4 minutes of data from the two
> >> inputs.
>
> >> way in which a shell gets moved by a sculling stroke. To a greater or
> >> lesser extent, every single recorded stroke from every boat in my data
> >> exhibits it. Until yesterday, my thoughts were that perhaps it was
> >> pry-bar” as Zeke put it recently; or perhaps it was some resonance or
> >> “bounce-back” of the oar-shaft after the initial catch.
>
> >> I went to see Carl Douglas a couple of days ago at his factory to get
> >> some seats repaired but also to show him the toy and some of the data
> >> collected so far. He pointed to a Rowperfect sitting nearby and
> >> suggested sticking an accelerometer on that. So I did, and do you know
> >> Rowperfect too. I’ve posted a couple of images atwww.slidingseat.net
>
> >> This, and the fact that the Rowperfect signal is in all other respects
> >> very similar to the sculling signal, suggests a couple of things to
> >> me:
>
> >> 1 The Rowperfect is a not-bad simulator for rowing/sculling
>
> >> 2a The post-catch acceleration hump/dip may be bio-mechanical rather
> >> than an artefact of rigging/materials
>
> >> 2b The dip may still be an artefact of rigging/materials, but
> >> instilled into my muscle-memory by 27 years of rowing such that I
> URLs: www.carldouglas.co.uk(boats) &www.aerowing.co.uk(riggers)
Carl asks: in what way are he data . . . meaningless?
I didn't say they were meaningless. I said the rowing head
acceleration data were meaningless *without corresponding center-of-
mass acceleration data.*
Nevertheless, I withdraw the "meaningless" as arguably pejorative.
Incomplete, then.
In a system where the shell/rowing head is coupled to a much more
massive rower or set of rowers, it should be reasonably easy to mess
with the light part's acceleration by doing something of otherwise
limited meaningfulness with the heavy part. If, to take my guess as an
example, for most of the stroke the body c of g moves somewhat less w
r t the shell than the inner ends of the oars do, but at a particular
place the torso and head are swinging over faster than the oar
handles, then entering that phase the body mass is accelerating bow-
wards w r t the shell and therefore the shell is accelerating stern-
wards w r t the body. Include that in the acceleration of the whole
system caused by the oars' interaction with the H2O and you would
observe a dip in shell acceleration. Measure the acceleration of the
shell-and-body system and there would be no such dip, assuming smooth
application of force and nothing odd happening between the blades and
the water . . . large assumptions, granted.//Zeke
Carl Douglas
> Many of you may have seen a short paper by Ken Young in 1991 entitled
> http://neutrino.phys.washington.edu/~wilkes/post/temp/phys208/shell.acceleration.html
> . Several times he draws attention to a common feature of many or most
> acceleration at 70-90 degrees (relative to the forward direction)�. He
> speculates at the end that it might be symptomatic of the blade
> stalling and suggests experimenting with different rigs etc to try to
> pin it down.
>
> Meantime, I got thinking on reading the paper that I could (get a
> friend to) build one of those and do some experimentation for myself.
> I did, and I now have data for several scullers (and a 2x) doing
> various different types of rowing.
>
> thumbnail-sized +/-2g inertial accelerometers. Generally I put one on
> the shell and one on the seat. One benefit of this arrangement is
> integrating for relative velocity, it is possible to determine the
> exact moment at which seat velocity is zero: i.e. frontstops. The
> datalogger has enough memory for 4 minutes of data from the two
> inputs.
>
> way in which a shell gets moved by a sculling stroke. To a greater or
> lesser extent, every single recorded stroke from every boat in my data
> exhibits it. Until yesterday, my thoughts were that perhaps it was
> pry-bar� as Zeke put it recently; or perhaps it was some resonance or
> �bounce-back� of the oar-shaft after the initial catch.
>
> I went to see Carl Douglas a couple of days ago at his factory to get
> some seats repaired but also to show him the toy and some of the data
> collected so far. He pointed to a Rowperfect sitting nearby and
> suggested sticking an accelerometer on that. So I did, and do you know
> Rowperfect too. I�ve posted a couple of images at www.slidingseat.net
>
> This, and the fact that the Rowperfect signal is in all other respects
> very similar to the sculling signal, suggests a couple of things to
> me:
>
> 1 The Rowperfect is a not-bad simulator for rowing/sculling
>
> 2a The post-catch acceleration hump/dip may be bio-mechanical rather
> than an artefact of rigging/materials
>
> 2b The dip may still be an artefact of rigging/materials, but
> instilled into my muscle-memory by 27 years of rowing such that I
> my body, and I play it back on the simulator)
>
> Comments appreciated
>
> Cheers, Magnus
I received the following today from my good friend, Cas Rekers:
I tried to send a message as a reply to the newsgroup, but failed to
sign in for one reason or another.
Would you please be so kind as to send my response below to the newsgroup ?
Quote:
Hello Magnus,
There is nothing mysterious about this dip, and the very interesting
accelerations graphs you show enable me to explain this clearly I hope.
The acceleration and the deceleration of the sculling shell and of the
Rowperfect sliding head are the resultant of the
acceleration/deceleration of the centers of mass of the rower and the
sculling_shell / Rowperfect_sliding_head relative to their common center
of mass, superimposed on the acceleration/deceleration of their common
center of mass relative to the environment. As the geometry of the body
changes during the translation from the catch to the finish position, so
does the position of the center of mass of the body relative to the
stretcher. The way in which this occurs is the result of the geometry of
the body and the division of mass between the various parts of the body,
and their pattern of movement. Theoretically one could calculate this
shift in position of the center of mass of the body if all these
parameters were known. To do so however is a very complicated .(Volker
Nolte in his thesis "The efficiency of the rowing stroke" has tried to
do so from images of a high speed camera)
On the Rowperfect the common center of mass of (rower + sliding head) is
at rest. Other than the minor friction forces of the sliding mechanism
and the seat and small gravity induced forces of the curved bar, there
are no external horizontal forces of any importance acting on the system
(Rower+sliding head). The laws of physics teach us that in that case the
relative horizontal movements of the centers of mass of the rower and of
the sliding head are inversely proportional to their masses.
So what you actually have done by measuring the acceleration and
deceleration of the sliding head of the Rowperfect is measuring this
very complicated shift in horizontal position of the center of mass of
your body relative to the stretcher during the rowing stroke cycle.
This pattern can be considered as your personal handwriting or
"rowprint"; partly originating from the dimensions of your body, and for
the other part depending upon the ingrained pattern of coordination of
the rowing stroke that you have acquired by training.
Other than on the Rowperfect, the common center of mass of you and your
boat is not at rest (otherwise you would never get at the finish) .At
any moment during the rowing stroke cycle this common center of mass is
accelerating when the sum of the total external horizontal forces in the
direction of movement is positive, and is decelerating when this sum is
negative. The actual sculling shell acceleration you measured is
therefore the resultant of the acceleration/deceleration of the center
of mass of your body relative to your boat ( your "rowprint") being
superimposed on the acceleration/deceleration of the common center of
mass of (you and your boat) relative to mother earth, due to the net
propulsive forces acting on the system.
My conclusion is that your suggestion 2.a is the correct one.
I take it that your suggestion 1 is a typical example of a British
understatement, and that you actually mean that the Rowperfect is a
jolly good simulator for rowing/sculling.
Cas Rekers,
Inventor of the Rowperfect.
(Ends Cas's response)
Carl
--
Carl Douglas Racing Shells -
Fine Small-Boats/AeRoWing Low-drag Riggers/Advanced Accessories
Write: Harris Boatyard, Laleham Reach, Chertsey KT16 8RP, UK
Find: http://tinyurl.com/2tqujf
magnus....@googlemail.com
> understatement, and that you actually mean that the Rowperfect is a
> jolly good simulator for rowing/sculling.
Cas, Yes you're quite right - my first draft of the posting contained
the words "very good", but whenever, at school, I used the word "very"
in my writing my teacher would tell me to take it out. As for how the
"good" changed to "not-bad" I'm not sure whether that's my Englishness
or my Finnishness. I'll stand by my original instinct: I think it's a
very good simulator.
At my club, Molesey, we have a couple of oldish RPs lying around, and
this weekend I'll try to do a more careful session with a warmed-up
rower rowing at race pressure and using both accelerometers: one on
the seat and one on the head. I'll then update the picture on my
little website.
Magnus
Leo
I have posted a few graphs in a pdf document on my www site to
illustrate how I try to model the "mysterious" bumps and hollows
in acceleration curves. See:
http://www.cyberiad.net/rsr1.pdf
Measured data come from a trial in 2008 using a Sykes pair
crewed by Olympians Drew Ginn and Duncan Free and measured by the
Aust. Inst. of Sport. (I've used data for 5 strokes at 36.1spm for
sake of this example).
The first page shows body segment angle regimes and the trajectories
of the segment centres of mass. The rowers' actual anthropometric
data was used to estimate the trajectories.
We also need to know the oar angle regimes and the forces exerted by
the rowers. Good measurements are usually available for these
quantities as shown on the second page.
Then we need to know the hydrodynamic forces and moments on the hull
that induce sinkage and trim (aka "squat") because these change the
underwater shape of the hull. The location of the bow and stern of
the hull during a single stroke are shown in the top plot on the
third
page. These curves (red bow, green stern) were calculated using the
combined effect of the rowers' centres of mass and the hull squat.
Of course, the results depend on the location in the shell of the two
crew members and the exact shape and proportions of the shell
(Incidentally, Ginn and Free are a bit "mismatched", inasmuch as Ginn
weighed about 87kg, and Free about 98kg.)
Once we have all that, we can estimate the hydrodynamic drag as shown
at the bottom of the third page. We must also burn some offerings
to the fluid gods and then make a few assumptions about air drag
because it is such a bloody complicated flow. After collecting some
empirical drag coefficients for the riggers, oars, human bodies and
hull we might get a reasonable estimate. Or maybe not: air gods can
be fickle.
Having assembled that morass, we can now estimate the instantaneous
hull acceleration and velocity as shown on the last page.
As can be seen in the top graph, there are all manner of bumps
and valleys in both the measurements and the predictions, and these
can sometimes be traced back to the influence of one or more of the
factors illustrated in the previous pages.
Finally, with a good set of experimental data and a reasonable model,
we can make changes to one or more factors (e.g. rigging, crew
placement etc) and see how they affect the acceleration curve.
If you can make it to the 7th International Symposium on Computer
Science in Sport (IACSS 2009) in Canberra from 22nd to 25th Sept.,
I'll show you how to do it!
http://www.ausport.gov.au/about/events/iacss2009/iacss2009
All the best,
Leo Lazauskas.
P.S. Thank you for this opportunity to be civil to Englishmen
before the Test series starts.
Mike De Petris
> Esteemed Collaborators,
> I have posted a few graphs in a pdf document on my www site to
> illustrate how I try to model the "mysterious" bumps and hollows
> in acceleration curves. See:http://www.cyberiad.net/rsr1.pdf
>
> Measured data come from a trial in 2008 using a Sykes pair
> crewed by Olympians Drew Ginn and Duncan Free and measured by the
> Aust. Inst. of Sport. (I've used data for 5 strokes at 36.1spm for
> sake of this example).
Do graphs on page 2 mean that Free is doing a shorter but stronger
stroke then Ginn?
Leo
Yes, slightly shorter stroke and a significantly higher peak axial
force.
Mike De Petris
> > Do graphs on page 2 mean that Free is doing a shorter but stronger
> > stroke then Ginn?
>
> Yes, slightly shorter stroke and a significantly higher peak axial
> force.
what a magic is rowing a pair B->
kc
What are you defining as "axial force" Which direction along the oar is
this? If I were to use the word "axial" in any way describing something
about an oar, or happening to an oar, I would assume "axial" meant
"along the long axis of the oar".... i.e. the direction pointing from
the handle to the blade.
Given your graphs, I'm assuming you are using axial to mean something
quite different? Like maybe the force perpendicular to the handle?
(i.e. pulling force?)
Or not?
Thanks for some clarification,
KC
kc
Indeed. It's my favorite boat to row.
Interesting how similar the force profiles of bow & stroke are in the
EARLY part of the stroke. Only after the first 1/3 of the stroke do
their force profiles start to diverge significantly. Note also that
this divergence then, occurs mostly during the stalled-blade phase of
the stroke.
Given what it takes to make a straight pair go straight, this makes
sense to me. I wonder what the rudder was doing during these strokes?
Interesting stuff. :-)
-KC
kc
You state that the athlete's own anthropometric data were used to
estimate trajectories of the segmental CoM's. One can't really
determine the segmental CoM's without dissecting the individual, so I
assume you estimated these. Just wondering, did you use DaLeva's
suggestions for inertial estimations or someone else?
-KC
Leo
> Great stuff, Leo. Thanks for sharing.
>
> You state that the athlete's own anthropometric data were used to
> estimate trajectories of the segmental CoM's. One can't really
> determine the segmental CoM's without dissecting the individual, so I
> assume you estimated these. Just wondering, did you use DaLeva's
> suggestions for inertial estimations or someone else?
>
Yes, I used de Leva's suggestions in conjunction with the rowers' body
segment lengths. I do have the rowers' segment volumes, but I haven't
been bothered to convert them to masses yet.
And, yes, I use the confusing term "axial force" to denote that the
force is along the axis of the shell, i.e. not along the oar length.
Leo
They might have produced slightly different force patterns during the
Olympic races. Ginn, I believe, had serious back trouble, so the
stroke lengths and forces during those races might have been
different.
kc
> On Jun 14, 10:27 am, kc <kc_n...@sonic.net> wrote:
>> Great stuff, Leo. Thanks for sharing.
>>
>> You state that the athlete's own anthropometric data were used to
>> estimate trajectories of the segmental CoM's. One can't really
>> determine the segmental CoM's without dissecting the individual, so I
>> assume you estimated these. Just wondering, did you use DaLeva's
>> suggestions for inertial estimations or someone else?
>>
>
> Yes, I used de Leva's suggestions in conjunction with the rowers' body
> segment lengths. I do have the rowers' segment volumes, but I haven't
> been bothered to convert them to masses yet.
Impressive that your predictions are so close to actual measured
accelerations. In our academic sports/gait lab (force plates in the
floor) we often show our students that predicting acceleration based on
kinematics doesn't match up well with acceleration based on the measured
force, as an example of how even the best kinematic data is somewhat
less than perfect.
> And, yes, I use the confusing term "axial force" to denote that the
> force is along the axis of the shell, i.e. not along the oar length.
I take it your transducers are on the oarlock pin, and only report force
in that one direction? Kind of unfortunate, but common. I don't know
why the few commercial products out there for gathering force data at
the oarlock pin do this. I would think most scientists would be happy
to have as many components of the actual force on the pin as possible.
I was wondering about the joint angle graphs and the statement "assumed"
referring to the joint excursion paths... did you not measure any
kinematics?
-KC
Leo
> Leo wrote:
>
> > And, yes, I use the confusing term "axial force" to denote that the
> > force is along the axis of the shell, i.e. not along the oar length.
>
> I take it your transducers are on the oarlock pin, and only report force
> in that one direction?
No, I have more but I just showed the force in the x-direction. I also
left out several other key factors, e.g. the behaviour of oar lift and
drag.
I showed my series of graphs to illustrate the point that the
existence of unusual bumps and valleys in acceleration curves can be
caused by one or more (interacting) factors and not just blade stall.
> I was wondering about the joint angle graphs and the statement "assumed"
> referring to the joint excursion paths... did you not measure any
> kinematics?
Only very coarsely myself. There's a lot being done on that at the
moment.
Cheers,
Leo.