An Artificial Neural Network Approach for Solving Inverse Kinematics Problem for an Anthropomorphic Manipulator of Robot SAR-401
Round 1
Reviewer 1 Report
In this paper, authors have soveld the IKP for the robot SAR-401 anthropomorphic manipulator. To solve the problem, the NN was designed, the input data of the NN are the elements of the homogeneous tranformation matrix. The choice of the NN configuration was carried out using a comparative anlysis. The solution obtained in this paper is characterized by high speed and accuracy. Given the presence of two manipulators in the robot SAR-401, when performing technological operations, self-collisions of manipulators may occur.
Author Response
The authors thank the reviewer for the positive review
Reviewer 2 Report
The paper presents an artificial neural network approach for solving inverse kinematics problem for an anthropomorphic manipulator of Robot SAR-401.
The paper is interesting and well written however there are some aspects that may improve the quality of the presentation:
Please insert measuring units on each graph. On figures 19-22 what is the value of 400 is it seconds (if so isn’t 400 second o little bit to slow for the robot?)
Can you provide previous studies in the domain to conclude if the obtained error using the correctional NN is an allowable one?
Author Response
Response to Reviewer 2 Comments
Point 1: «Please insert measuring units on each graph. On figures 19-22 what is the value of 400 is it seconds (if so isn’t 400 second o little bit to slow for the robot?)».
Response 1: Figures corrected.
Figures 19-22 show the points along the abscissa axis as described on page 18. Appropriate clarifications have been made in these figures. The time of movement of the manipulator to the specified points is determined by setting the velocity of movement in the control program, taking into account the limitations of the servos.
Point 2: «Can you provide previous studies in the domain to conclude if the obtained error using the correctional NN is an allowable one?».
Response 2: Corrected
Calculation accuracy and neural network (NN) characteristics are shown in Table 5.. The accuracy of 2-4 mm obtained by correctional NN is sufficient for the manipulation tasks of the robot SAR-401 and is related to mechanical characteristics (backlash, friction, etc.). The specified error allows robot to capture objects and perform manipulation operations.
This information is added on page 22 after Figure 22.
Author Response File: 
Author Response.pdf
