A Novel FastSLAM Framework Based on 2D Lidar for Autonomous Mobile Robot

Round 1

Reviewer 1 Report

Equation 1 is not clear. In the left part of the equation, a opening parenthesis is missing, because the right part, first term, is equal than the left part; well, the equation is not clear at all, and must reformulated.

Equation 2 has also some problems, the term in the denomiantor (pi) is not defined, and the position of the superindex i is not clear, comapred with the same term in the numerator, please correct the equation and define the term pi.

Equation 3 is not clear; the threshold is Ne, but the equation does not calculate Ne, instead Neff; but at the same time authors said that Neff is the number of particles used as a threshold. Well, the equation and explantion is not clear at all. Please, reformulate the equation and name the variables correctly.

Equation 6, it is missed a parenthesis in w(i)t > Wt

Equation 7, the superindex of the left part is not i, because the number of particles is the sum of xL + xH, then a new variable (eg. k) should be added, in the terms of k = i+j

In fact, the new proposal for an adaptive GA resamplig is based on the distribution of 1/4 in the variance, with the unique justification of paper [26]. The reasoning is very weak though.

In equation 11, the subindex id is not clear, what is the meaning of id?
In equation 12, to update the position xt+1 is calculated with the sum of a position xt plus a velocity; well in terms of physics, the velocit should be converted to space multiplying by the sample period; in this case this fact should be indicated somehow.

In general, the approach requires the adjustment of many paramentres (sigma, fractional differentiation ordre,...) which authors took from other studies, but maybe when all is out together the readjustment would yield better results. The fact that this possibility exists is pitfall in the overall proposal.

In simulation, real data should be used, for instance 3m/s is quite high speed for slam, and 20 rad/s is an unrealistic angular speed for any moving body.
Why authors say "the maximum measuring distance is 30 m" when the plots show larger distances than 30m?.

In my opinion the experiment in section 3.2, with the real car is invalid, because the groundtruth is calculated with an onboard GPS, and the position given by those devices contains large deviations, and therefore the error calculation needs another term based on GPS error, which is unknown and random in a specific range (also unknown). I would delete this subsection.
Instead the experiment in section 3.3 is more interesting, at 0.3 ms/s, one tenth the speed in the simulation experiment.

Clearly, it is missed a quantitative evaluation of the error among different methods, figure 13 is nice but needs a table with numerical values for the error.

In the table of symbols I missed f (eq 18) and pi (eq 2).

Ref 8 is cited in the text as Manuel[8], clearly should be Cugliari[8] instead.

Typos:
line 109, ith, the h is subindex
line 130, it is Retrieval
line 3 algorithm 1, mth, h subindex; in the rest of the algorithm appears as m_th, be coherent along the paper

 

Author Response

Dear Reviewer,

 

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 2 Report

Nice work! Figure 2 is a bit unclear in its way of representation and Fig 6 figures should be bigger as there is interest in the details.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

All my questions have been answered, and the paper can be published as it is