Feedback from Peter Stahlecker

[moorepants]

1.

Your trials, and trials before were (naturally) done on treadmills.

Is it obvious to the reader of such a paper that the dynamics studied are unaffected by the fact that the bike as a whole is at rest, just

the wheels are turning?

(Not obvious to me, but the intended audience matters.)

2,

I presume to a reader of this paper the terms roll angle, capsize, weave and wobble

have clear meanings.

3.

The writing in figure 3 is very small – but of course most reader will be much younger, with better eyes, than me.

4.

On page 5, near the bottom of the page is this sentence:

A recovery from the perturbation…..if a person neither places their foot on the treadmill surface, or their….

Should this not be nor?

[moorepants]

More:

1.

I assume, that the linear model, eq (1) is adequate, since if the roll angle is so large that it is no longer

adequate, the guy will fall off the bike anyway.

2.

On page 2, right above 2.5. Measurements is this sentence:

We randomize whether… during the first or second set of perturbations…

Should this not be and, I assume you randomize the on/off state of the assist in both sets?

(But my native language is German, not English)

3.

Table 1:

I am unclear, what the “perturbation number” is.

Did I overlook its definition?

4.

I assume, that steer angle and roll angle are independent because a perturbation is initiated at random, without looking

at the position of the bike.

5.

Eq (5) states that f_ij | p_ij ~ Bernoulli(p_ij)

Is this obvious for the intended audience?

(took me a while until I thought I understood – but I am not the intended audience, there are better educated)

6.

On page 8, about in the middle of the page there is this sentence:

We expected there to be a variation between participants in how well they are able to resist the perturbation.

I assume, that some of your 26 participants were ‘better’ at bike riding than others.

So, this sentence means that a ‘good’ biker rider and a ‘bad’ bike rider are equally affected by perturbations at slow speeds?

[moorepants]

I assume, that the linear model, eq (1) is adequate, since if the roll angle is so large that it is no longer adequate, the guy will fall off the bike anyway.

The linear model is adequate for predicting bicycle motion around the equilibrium (+/- 30 degs in steer and roll). For this paper, we are only interested if there is improved stability around the equilibrium and if that correlates to fewer falls.

[moorepants]

On page 2, right above 2.5. Measurements is this sentence:

We randomize whether… during the first or second set of perturbations…

Should this not be and, I assume you randomize the on/off state of the assist in both sets?

(But my native language is German, not English)

The current sentence is "We randomize whether the balance assist system is on or off during the first or second set of perturbations for each participant."

Maybe Peter thinks we need the word "state" in there. Not sure what he means.

[moorepants]

I assume, that steer angle and roll angle are independent because a perturbation is initiated at random, without looking at the position of the bike.

Yes.

[moorepants]

Eq (5) states that f_ij | p_ij ~ Bernoulli(p_ij)

It is a standard way to write such an equation, so I don't think we need to explain it more.

[moorepants]

So, this sentence means that a ‘good’ biker rider and a ‘bad’ bike rider are equally affected by perturbations at slow speeds?

Essentially yes.

[moorepants]

He meant:

What I meant was: Is it obvious to the audience, that f_ij given p_ij is Bernoulli with parameter p_ij?

So more like why should we choose this probability distribution.

[moorepants]

I reordered the wording around the Bernoulli equation, maybe it is a bit better but I didn't really explain it more, leaving it to the textbook definition elsewhere.