Online Predictive Visual Servo Control for Constrained Target Tracking of Fixed-Wing Unmanned Aerial Vehicles
Round 1
Reviewer 1 Report
Comments and Suggestions for AuthorsThe authors in this paper developed a procedure to achieve the target tracking for a fixed-wing UAV with a pan-tilt camera based on the method combining MPC and IBVS.
The following points are required to be addressed by the authors to make the paper clear and readable by others.
11. The authors pointed out in Line 102, that among their contributions is “Extensive experimental”. The word experimental should be removed since there is no experimental work.
22. Increase the size of Figure 1.
33. In Subsection 2.1, the authors assumed the UAV flies with constant speed and fixed height. The application of such a system is limited to searching for objects in the desert or in the sea, for example. This system cannot do any maneuvering to avoid obstacles. See for example (add them):
a. Modeling and simulation of quadrotor UAV with tilting rotors
b. Decentralized backstepping control of a quadrotor with tilted rotor under wind gusts
Any other applications for such a proposed system?
44. Provide a reference for the model in Equation 1.
55. The authors assumed that the camera is moving in two dimensions, but in Line 145 they assumed 3 dimensions. Explain?
66. In Equation 10, is there a derivative of s2? Use ∆t to denote the sampling time.
77. Is there a typo mistake in line 192, u(k-1) or u(k-1/k)?
88. Also, the is a typo mistake in line 197.
99. Over the text, no definition or reference for HITL.
Comments on the Quality of English LanguageNone
Author Response
Please see the attachment.
Author Response File: 
Author Response.pdf
Reviewer 2 Report
Comments and Suggestions for AuthorsThe paper considers a fixed-wing target tracking task deployed using predictive cotnrol framework. In the introductory part of the paper, the authors provide some insight into a span of various tracking (surveillance) tasks, supporting their choice. One of the means to do it, are vision techniques. Some of the approaches must be robust with respect to various possible errors, such as designing the controller using image plane data, which might be corrupted.
The citing style is proper, as the authors go one-by-one in their discussion. Lines 78-80 are not well understood, as the authors oppose real-time to on-line approaches. Could you clearly define the difference, please?
In the proposed approach the authors stabilize the tilt camera in the image so that the target is in its centre. THis might be hard during rapid movements of the target.
The contribution is clearly stated, but the novelty statement is missing though.
Section 2 - why is circle chosen, not e.g. a lemniscate shape when its centre could be always close to the target, and the displacement is reduced twice with respect to the tracked target?
To describe the camera dynamics, the authors provide a Jacobian approach. The question is why not to build a feedback linearization scheme here, as the model of the dynamics is well known, does not change over time, and should be easy to apply? Would that have any influence on the performance of the system?
Derivation of J1 - is it possible that alpha = pi/2? What is your approach then?
How do you tune Qs and Qu matrices? What is the impact of their diagonals on the overall results?
Reformulation of (26) carries significant problems with discontinuity of the penalty funcion component. Why not to use a squared value of a max function?
The proof of your solution is given, but no issues concerning lack of differentialibility of the max is not included.
IN the remaining part of the paper, the authors provide a full support for their solution, which is pretty impressing and I have no comments on it.
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
Reviewer 3 Report
Comments and Suggestions for AuthorsThis paper proposes an online predictive control method for fixed-wing unmanned aerial vehicles (UAVs) with a pan-tilt camera in target tracking.
[21] is not handled in the introduction.
[4] considers vision-based detection and tracking of a mobile ground target using a fixed-wing UAV. can you compare your method with [4] using experiments?
computation load of the proposed method needs to be compared with that of the state of the art methods. is it possible to run your method in real time?
can you compare the performance of your method with the state of the arts methods, in the case where you track a target with nonlinear motion?
can you compare the performance of your method with the state of the arts methods, in the case where you track a target whose trajectory is long and complicated? consider a target which changes its speed and yaw.
Comments on the Quality of English Languageok
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
Round 2
Reviewer 3 Report
Comments and Suggestions for AuthorsI am satisfied with revision.
Comments on the Quality of English Languageok
