Software package for the study of neural network technologies as part of a direct spherical drive control system

Authors: Ageev A.V.
Published in issue: #4(45)/2020
DOI: 10.18698/2541-8009-2020-4-596
Category: Informatics, Computer Engineering and Control | Chapter: System Analysis, Control, and Information Processing, Statistics
Keywords: neural network, automatic control, computer simulation, mobile robot, Robotic Operating System, MATLAB, Gazebo, Tensorflow
Published: 21.04.2020

The purpose of the study is the formation of a software environment for studying and modeling a control system for a direct spherical drive (DSD) using a neural network architecture. Modeling systems MATLAB, Gazebo and the neural network library Tensorflow are considered as tool software. A feature of the control system modeling is that this computer model should be able to integrate into the modeling systems of other parts of an autonomous mobile robot. To solve the integration problem, the Robotic Operating System (ROS) was used, which allows developing software modules for connecting systems that initially do not support this feature.

References

[1] Sokolov S.M., Trifonov O.V., Yaroshevskiy V.S. The control system of spherical direct drive actuator with Hall sensors in a feedback. Preprinty IPM im. M.V. Keldysha [KIAM Preprint], 2008, no. 94 (in Russ.).

[2] Klavdienko A.A., Manyukhina A.N. The review of simulation tools for bipedal robots. Nauka i Mir [Science and world], 2016, no. 4(32), pp. 52–58 (in Russ.).

[3] Gusev A.A., Molotov R.S. Implementation of interaction of combat vehicles models in the military simulator in Unity ide. Vestnik UlGTU, 2016, no. 3(75), pp. 44–47 (in Russ.).

[4] Mezhenin A.V. Virtual 3D medium as verification and test tool at designing. Prioritetnye nauchnye napravleniya: ot teorii k praktike, 2016, no. 21, pp. 105–110.

[5] Mikhaylova U.V., Mikhaylov E.A., Sarvarov A.S. Software solutions for the development of robot control system architecture. Elektrotekhnicheskie sistemy i kompleksy [Electrotechnical Systems and Complexes], 2013, no. 21, pp. 111–117 (in Russ.).

[6] Wang L., Yang Y., Correa G., et al. Fearing OpenRoACH: a durable open-source hexapedal platform with onboard robot operating system (ROS). ICRA, 2019, pp. 9466–9472. DOI: https://doi.org/10.1109/ICRA.2019.8794042

[7] Koenig N., Howard A. Design and use paradigms for gazebo, an open-source multi-robot simulator. IEEE/RSJ IROS, 2004, pp. 2149–2154. DOI: https://doi.org/10.1109/IROS.2004.1389727

[8] Gazebo parallel physics report. DARPA robotics challenge. OSRF, 2015.

[9] MLIR: A new intermediate representation and compiler framework. medium.com: website. URL: https://medium.com/tensorflow/mlir-a-new-intermediate-representation-and-compiler-framework-beba999ed18d (accessed: 10.11.2019).

[10] Skobelev M.M. Development of MATLAB-Simulink model of visualization in virtual reality of controlled spatial motion of underwater vehicle. Nauka i obrazovanie: nauchnoe izdanie [Science and Education: Scientific Publication], 2011, no. 10. URL: http://technomag.edu.ru/doc/234292.html (in Russ.).

[11] ROS. Stek navigatsii [ROS navigation stack]. habr.com: website (in Russ.). URL: https://habr.com/ru/post/327888/ (accessed: 14.12.2019).

[12] Robot-telezhka na ROS. Chast’ 3. Uskoryaemsya, menyaem kameru, ispravlyaem pokhodku [ROS trolley robot, Part 3. Accelerating, changing camera, improving motion]. habr.com: website (in Russ.). URL: https://habr.com/ru/post/463147/ (accessed: 14.12.2019).