Central pattern generators for biomorphic robotics
- Authors: Zharinov A.I.1,2, Potapov I.A.1, Kurganov D.V.1, Lobov S.A.1,2
-
Affiliations:
- National Research Lobachevsky State University of Nizhny Novgorod
- Immanuel Kant Baltic Federal University
- Issue: Vol 18, No 4 (2023)
- Pages: 748-751
- Section: Conference proceedings
- URL: https://genescells.ru/2313-1829/article/view/623314
- DOI: https://doi.org/10.17816/gc623314
- ID: 623314
Cite item
Open Access
Access granted
Subscription or Fee Access
Abstract
Typically, the structure of the robot fish frame significantly differs compared to the real organism. One significant difference is in the number of body segments. While live fish can have between 16 (moon fish) to 400 belt fish [1] segments, robots usually have only 5–6 segments since substantial precision is unnecessary when simulating movement. At the same time, this method limits a significant portion of the control circuit’s structure compared to a fish’s nervous system because it only requires control over a smaller number of body segments.
Control systems using different oscillators can simulate the functioning of fish central generators [2–4]. Typically, each half-center of the fish’s CPGs is interconnected with and mutually inhibitory towards the others, with each being responsible for the antagonist muscles. In this case, the generator’s pattern characteristics stem from the mutual influence of oscillator-antagonist pairs connected to each other. The half-centers’ interaction mechanism with each other is designed to match the movement pattern’s desired final parameters.
This “artificial” approach is unsuitable for working with spike neurons because the mechanisms of cellular interaction are well-defined. Altering how cells interact with each other when creating a biologically relevant model is also undesired. Here we present evidence that incorporating select physiological traits of fish into the design of a CPG structure utilizing spike neurons can enhance the system’s functional capacity.
Previously, we demonstrated a half-center CPG model using Izhikevich neurons [5]. This model can serve as a control loop for a tuna robot. Although this development aligns with the fundamental principles of CPG organization in fish, reproducing the typical generator mode of operation for pike on it proved challenging. This is due to the fact that the anguliform type of locomotion implies the presence of a moving wave, which means a phase lag in the activation of half-centers.
One potential solution to the problem lies in the physiology of fish, specifically the structure of their muscle fibers. Fish have muscle segments called myomeres, which correlate with the number of vertebrae and spinal centers that create the CPG. A distinguishing feature of these myomeres is their zigzag shape. To achieve body bending at a single point, it requires a collaborative action between multiple myomeres and corresponding CPG segments.
While our model assumes control of the entire fish with only 5 segments of the CPG, the actual pike includes 56–65 segments. To attain the necessary difference in activation phase between generator segments, we propose increasing the number of generator nodes responsible for operating a single propulsion unit.
Indeed, increasing the number of transmission segments resulted in a steady divergence in activation phases among successive segments of the CPG.
However, the incorporation of supplementary segments fails to address the challenge of managing the frequency of the CPG’s operation and shifting between patterns. Consequently, we intend to integrate CPG neurons of diverse types outlined in the model, along with introducing feedback to rectify its modes of operation in the future.
Full Text
Typically, the structure of the robot fish frame significantly differs compared to the real organism. One significant difference is in the number of body segments. While live fish can have between 16 (moon fish) to 400 belt fish [1] segments, robots usually have only 5–6 segments since substantial precision is unnecessary when simulating movement. At the same time, this method limits a significant portion of the control circuit’s structure compared to a fish’s nervous system because it only requires control over a smaller number of body segments.
Control systems using different oscillators can simulate the functioning of fish central generators [2–4]. Typically, each half-center of the fish’s CPGs is interconnected with and mutually inhibitory towards the others, with each being responsible for the antagonist muscles. In this case, the generator’s pattern characteristics stem from the mutual influence of oscillator-antagonist pairs connected to each other. The half-centers’ interaction mechanism with each other is designed to match the movement pattern’s desired final parameters.
This “artificial” approach is unsuitable for working with spike neurons because the mechanisms of cellular interaction are well-defined. Altering how cells interact with each other when creating a biologically relevant model is also undesired. Here we present evidence that incorporating select physiological traits of fish into the design of a CPG structure utilizing spike neurons can enhance the system’s functional capacity.
Previously, we demonstrated a half-center CPG model using Izhikevich neurons [5]. This model can serve as a control loop for a tuna robot. Although this development aligns with the fundamental principles of CPG organization in fish, reproducing the typical generator mode of operation for pike on it proved challenging. This is due to the fact that the anguliform type of locomotion implies the presence of a moving wave, which means a phase lag in the activation of half-centers.
One potential solution to the problem lies in the physiology of fish, specifically the structure of their muscle fibers. Fish have muscle segments called myomeres, which correlate with the number of vertebrae and spinal centers that create the CPG. A distinguishing feature of these myomeres is their zigzag shape. To achieve body bending at a single point, it requires a collaborative action between multiple myomeres and corresponding CPG segments.
While our model assumes control of the entire fish with only 5 segments of the CPG, the actual pike includes 56–65 segments. To attain the necessary difference in activation phase between generator segments, we propose increasing the number of generator nodes responsible for operating a single propulsion unit.
Indeed, increasing the number of transmission segments resulted in a steady divergence in activation phases among successive segments of the CPG.
However, the incorporation of supplementary segments fails to address the challenge of managing the frequency of the CPG’s operation and shifting between patterns. Consequently, we intend to integrate CPG neurons of diverse types outlined in the model, along with introducing feedback to rectify its modes of operation in the future.
ADDITIONAL INFORMATION
Funding sources. This work was supported by the Russian Science Foundation (project No. 21-12-00246).
About the authors
A. I. Zharinov
National Research Lobachevsky State University of Nizhny Novgorod; Immanuel Kant Baltic Federal University
Author for correspondence.
Email: zharinov@neuro.nnov.ru
Russian Federation, Nizhny Novgorod; Kaliningrad
I. A. Potapov
National Research Lobachevsky State University of Nizhny Novgorod
Email: zharinov@neuro.nnov.ru
Russian Federation, Nizhny Novgorod
D. V. Kurganov
National Research Lobachevsky State University of Nizhny Novgorod
Email: zharinov@neuro.nnov.ru
Russian Federation, Nizhny Novgorod
S. A. Lobov
National Research Lobachevsky State University of Nizhny Novgorod; Immanuel Kant Baltic Federal University
Email: zharinov@neuro.nnov.ru
Russian Federation, Nizhny Novgorod; Kaliningrad
References
- Suharenko EV, Maksimov VI. Fiziologiya ryb. Kerch’: KSMTU; 2021. 156 p. (In Russ).
- Wang W, Guo J, Wang Z, Xie G. Neural controller for swimming modes and gait transition on an ostraciiform fish robot. In: 2013 IEEE/ASME International Conference on Advanced Intelligent Mechatronics (AIM), 2013; Wollongong, Australia. Jul 9–12; 2013:1564–1569.
- Yu J, Wang C, Xie G. Coordination of Multiple Robotic Fish with Applications to Underwater Robot Competition. IEEE Transactions on Industrial Electronics. 2016;63(2):1280–1288. doi: 10.1109/TIE.2015.2425359
- Bal C, Ozmen Koca G, Korkmaz D, et al. CPG-based autonomous swimming control for multi-tasks of a biomimetic robotic fish. Ocean Engineering. 2019;189:106334. doi: 10.1016/j.oceaneng.2019.106334
- Zharinov AI, Kurganov DP, Potapov IA, et al. Self-organizing CPGs in the control loop of a biomorphic fish robot. In: 2022 Fourth International Conference Neurotechnologies and Neurointerfaces (CNN), 2022; Kaliningrad, Russia. Sept 14–16; 2022:219–222. doi: 10.1109/CNN56452.2022.9912568
Supplementary files
