Secord et al. review:

Overall, the paper covered interesting and novel work and presented a logical progression of dynamical modeling, system identification, and experimental results from parameter tuning.  However, some minor grammatical, clarity, and citation changes could improve the paper.  

Specific comments by page

Pg 1:

Raiberts and Cutkoskys require apostrophes.

"If an actuator possesses tunable compliance inherent in its structure, robots may be more efficient and dexterous" - No reason or citation given

Is 20% displacement mentioned in the introduction a maximum, or is it the amount chosen for your device?  Couldn't obtain reference 18 to check this.  *Further reading implies it to either be a number chosen to mimic natural muscle strain, or else a number limited by "practical design requirements" as mentioned on pg 2.*

Pg 2:

"This type of amplification is a standard technique in the literature."  Citation?

"As shown, the cellular architecture allows the PZT actuator to be used in a way similar to the anatomical structure of animal muscles" Further explanation is necessary

"This choice is justified since experimental data indicate that the dynamics of the first layer units are unimodal with a natural frequency well outside the bandwidth of the second layer."  No data or citation given, but perhaps not needed.  

Pg  3:

In eqn 2, the term 'b' is not defined.

Under kinematics of the second layer chain, notation is confusing, with 'i' and 'j' referring both to a unit vector and to unit/link number

Pg 4: 

Is the bronze rhombus commercially available or manufactured in house?  Is the flexure material also bronze (ie. a leaf spring?), or is a softer joint material used?  

Pg 5:

Under discussion of model competence, it isn't clear why the large disparities (due to damping) won't affect control.  While pole-zero topology is correctly predicted, tuning for resonance requires close agreement between actual and theoretical.   A sentence or two here might help.  *Further reading revealed that the wing dynamics were dominant in the system, as the errors from pg 7 are much lower.*

Pg 7:

In fig 11, half-wing mass is difficult to read on the plot

The word reasonablent appears between eqn 18 and eqn 19

Minor errors such as "thickness of width parameters"