Методика построения гибридной нейросистемы реального времени для решения обратной задачи кинематики избыточного манипулятора

Павел [Pavel] Евгеньевич [E.] Ганин [Ganin]; Александр [Aleksandr] Исаакович [I.] Кобрин [Kobrin]

doi:10.24160/1993-6982-2018-4-128-137

Павел [Pavel] Евгеньевич [E.] Ганин [Ganin]
Александр [Aleksandr] Исаакович [I.] Кобрин [Kobrin]

DOI: https://doi.org/10.24160/1993-6982-2018-4-128-137

Keywords: inverse kinematics problem, robotic manipulator, real-time control system, neural network

Abstract

Solution of the inverse kinematics (IK) problem with the aid of neural networks is considered. The control system for a multi-link redundant manipulator is synthesized. The configuration of a control system on the basis of servo drives is presented. The proposed control system is based on an algorithm involving a new hybrid method for solving the IK problem. This method combines the ANFIS adaptive neural fuzzy inference network and an iterative refinement algorithm (according to the Newton-Raphson method). Thus, the proposed algorithm combines the advantages of the neural network and iterative approaches, namely, high precision and high response speed. The required coordinates (link rotation angles) for the inverse problem are calculated in the neural network (ANFIS), after which they are refined by the iteration method. Hence, a much fewer number of iterations have to be carried out in the numerical method, and, accordingly, much shorter time is taken to execute the algorithm. The developed algorithm ensures a controlled accuracy of calculations with due regard of its application in real-time control systems. The developed control system is universal in nature and can be used for manipulators having different design parameters. For investigating the control system synthesized proceeding from the developed method for solving the IK problem, a three-link manipulator design was considered. The mathematical description used for constructing the manipulator’s operating space and for training the control system neural network is presented. The results from experimental investigations of applying the hybrid algorithm for calculating the link and actuator coordinates are given. The method was investigated in the Matlab environment. The results of the performed experiments allowed a conclusion to be drawn about the possibility of using the developed method in real-time control systems. To solve the problem of shaping the operating spaces of certain types of manipulators, a graphical interface was developed using the Matlab GUI. The application includes features for adjusting the design parameters (the mechanical structure and accuracy). As an example, the workspaces of both planar and spatial manipulator designs with the specified parameters are taken.

Information about authors

Павел [Pavel] Евгеньевич [E.] Ганин [Ganin]

Workplace

Control and Informatics Dept., NRU MPEI

Occupation

Assistant

Александр [Aleksandr] Исаакович [I.] Кобрин [Kobrin]

Science degree:

Dr.Sci. (Phys.-Math.)

Workplace

Robotics, Mechatronics, Machines Dynamics and Strength Dept., NRU MPEI

Occupation

Professor

References

1. Pieper D., Roth B. The Kinematics of Manipulators Under Computer Control // Proc. Second Intern. Congress Theory of Machines and Mechanisms. 1969. V. 2. Pp. 159—169.

2. Капустина О.М. Описание множества точных решений обратной задачи кинематики робота KUKA YOUBOT c помощью обобщенных координат его платформы // Естественные и технические науки. 2016. № 12 (102). С. 176—180.

3. Kofinas N., Orfanoudakis E., Lagoudakis M.G. Complete Analytical Inverse Kinematics for NAO //Proc. 13 th Intern. Conf. Autonomous Robot Syst. 2013. Pp. 1—6.

4. Barinka L. Inverse Kinematics — Basic Methods. [Электрон. ресурс] http://old.cescg.org/CESCG-2002/LBarinka (дата обращения 09.11.2017).

5. Corke P. Robotics, Vision and Control. Fundamental Algorithms In MATLAB ® . Springer, 2017. Рр. 245—263.

6. Ломовцева Е.И., Челноков Ю.Н. Дуальные матричные и бикватернионные методы решения прямой и обратной задач кинематики роботов-манипуляторов на примере стэнфордского манипулятора // Известия Саратовского ун-та. Серия «Математика. Механика. Информатика». 2014. Т. 14. № 1. C. 88—95.

7. Дыда А.А., Оськин Д.А. Решение обратной задачи кинематики для манипуляционного робота методом штрафных функций // Фундаментальные исследования. 2015. № 11-4. С. 673—677.

8. Binggul Z., Ertunc H., Oysu C. Comparison of Inverse Kinematics Solutions Using Neural Network for 6R Robot with Offset // Proc. Congress on Computational Intelligence Method and Appl. 2005. Pp. 1—5.

9. Driscoll J. Comparison of Neural Network Architectures for the Modeling of Robot Inverse Kinematics // IEEE Trans. on Computers. 2000. Pp. 44—51.

10. Shital S., Chiddarwar N., Ramesh B. Comparison of RBF and MLP Neural Networks to Solve Inverse Kinematic Problem for 6R Serial Robot by a Fusion Approach // Eng. Appl. of Artificial Intelligence. 2010. No. 23 (7). Pp. 1083—1092.

11. Ankarali A., Cilli M. ANFIS Inverse Kinematics and Hybrid Control of a Human Leg Gait Model // APJES I-II. 2013. Pp. 34—49.

12. Manjaree S., Agarwal V., Nakra B. Inverse Kinematics Using Neuro-Fuzzy Intelligent Technique for Robotic Manipulator // Intern. J. Advanced Computer Research. 2013. No. 3(4). Pp. 160—165.

13. Layatitdev D., Jajneswar N., Mahapatra S. A Comparative Study of Prediction of Inverse Kinematics Solution of 2-DOF, 3-DOF and 5-DOF Redundant Manipulators by ANFIS // Intern. J. Computer Sci. and Network. 2014. V. 3. No 5. Pp. 304—308.

14. Joong-Kyoo P. Inverse Kinematics Based on Fuzzy Logic and Neural Networks for the WAM-Titan II Teleoperation System [Электрон. ресурс] http://trace.tennessee.edu/utk_gradthes/186 (дата обращения 09.11.2017).

15. Panchanand J., Bibhuti B., Prakash S. Inverse Kinematic Solution of Robot Manipulator Using Hybrid Neural Network. // Intern. J. Materials Sci. and Eng. 2015. V. 3. No. 1. Pp. 31—38.
---
Для цитирования: Ганин П.Е., Кобрин А.И. Методика построения гибридной нейросистемы реального времени для решения обратной задачи кинематики избыточного манипулятора // Вестник МЭИ. 2018. № 4. С. 128—137. DOI: 10.24160/1993-6982-2018-4-128-137.
#
1. Pieper D., Roth B. The Kinematics of Manipulators Under Computer Control. Proc. Second Intern. Congress Theory of Machines and Mechanisms. 1969;2:159—169.

2. Kapustina O.M. Opisanie Mnozhestva Tochnyh Resheniy Obratnoy Zadachi Kinematiki Robota KUKA YOUBOT c Pomoshch'yu Obobshchennyh Koordinat Ego Platformy. Estestvennye i tekhnicheskie Nauki. 2016;12 (102):176—180. (in Russian).

3. Kofinas N., Orfanoudakis E., Lagoudakis M.G. Complete Analytical Inverse Kinematics for NAO. Proc. 13 th Intern. Conf. Autonomous Robot Syst. 2013:1—6.

4. Barinka L. Inverse Kinematics — Basic Methods. [Elektron. Resurs] http://old.cescg.org/CESCG-2002/ LBarinka (Data Obrashcheniya 09.11.2017).

5. Corke P. Robotics, Vision and Control. Fundamental Algorithms In MATLAB ® . Springer, 2017:245—263.

6. Lomovtseva E.I., Chelnokov Yu.N. Dual'nye Matrichnye i Bikvaternionnye Metody Resheniya Pryamoy i Obratnoy Zadach Kinematiki Robotov-manipulyatorov Na Primere Stenfordskogo Manipulyatora. Izvestiya Saratovskogo Un-ta. Seriya «Matematika. Mekhanika. Informatika». 2014;14;1:88—95. (in Russian).

7. Dyda A.A., Os'kin D.A. Reshenie Obratnoy Zadachi Kinematiki dlya Manipulyatsionnogo Robota Metodom Shtrafnyh Funktsiy. Fundamental'nye Issledovaniya. 2015;11-4:673—677. (in Russian).

8. Binggul Z., Ertunc H., Oysu C. Comparison of Inverse Kinematics Solutions Using Neural Network for 6R Robot with Offset. Proc. Congress on Computational Intelligence Method and Appl. 2005:1—5.

9. Driscoll J. Comparison of Neural Network Architectures for the Modeling of Robot Inverse Kinematics. IEEE Trans. on Computers. 2000:44—51.

10. Shital S., Chiddarwar N., Ramesh B. Comparison of RBF and MLP Neural Networks to Solve Inverse Kinematic Problem for 6R Serial Robot by a Fusion Approach. Eng. Appl. of Artificial Intelligence. 2010; 23 (7):1083—1092.

11. Ankarali A., Cilli M. ANFIS Inverse Kinematics and Hybrid Control of a Human Leg Gait Model. APJES I-II. 2013:34—49.

12. Manjaree S., Agarwal V., Nakra B. Inverse Kinematics Using Neuro-Fuzzy Intelligent Technique for Robotic Manipulator. Intern. J. Advanced Computer Research. 2013;3(4):160—165.

13. Layatitdev D., Jajneswar N., Mahapatra S. A Comparative Study of Prediction of Inverse Kinematics Solution of 2-DOF, 3-DOF and 5-DOF Redundant Manipulators by ANFIS. Intern. J. Computer Sci. and Network. 2014;3;5:304—308.

14. Joong-Kyoo P. Inverse Kinematics Based on Fuzzy Logic and Neural Networks for the WAM-Titan II Teleoperation System [Elektron. Resurs] http://trace.tennessee.edu/utk_gradthes/186 (Data Obrashcheniya 09.11.2017).

15. Panchanand J., Bibhuti B., Prakash S. Inverse Kinematic Solution of Robot Manipulator Using Hybrid Neural Network.. Intern. J. Materials Sci. and Eng. 2015;3;1:31—38.
---
For citation: Ganin P.E., Kobrin A.I. A Procedure for Synthesizing a Hybrid Real-time Neural System for Solving a Redundant Manipulator’s Inverse Kinematics Problem. MPEI Vestnik. 2018;4:128—137. (in Russian). DOI: 10.24160/1993-6982-2018-4-128-137.