Spiking Neural Network Discovers Energy-Efficient Hexapod Motion in Deep Reinforcement Learning

In Deep Reinforcement Learning (DRL) for robotics application, it is important to find energy-efficient motions. For this purpose, a standard method is to set an action penalty in the reward to find the optimal motion considering the energy expenditure. This method is widely used for the simplicity...

Descripción completa

Guardado en:
Detalles Bibliográficos
Autores principales: Katsumi Naya, Kyo Kutsuzawa, Dai Owaki, Mitsuhiro Hayashibe
Formato: article
Lenguaje:EN
Publicado: IEEE 2021
Materias:
Acceso en línea:https://doaj.org/article/335681a7613746e5a4b027d670d343bf
Etiquetas: Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
id oai:doaj.org-article:335681a7613746e5a4b027d670d343bf
record_format dspace
spelling oai:doaj.org-article:335681a7613746e5a4b027d670d343bf2021-11-18T00:08:52ZSpiking Neural Network Discovers Energy-Efficient Hexapod Motion in Deep Reinforcement Learning2169-353610.1109/ACCESS.2021.3126311https://doaj.org/article/335681a7613746e5a4b027d670d343bf2021-01-01T00:00:00Zhttps://ieeexplore.ieee.org/document/9606760/https://doaj.org/toc/2169-3536In Deep Reinforcement Learning (DRL) for robotics application, it is important to find energy-efficient motions. For this purpose, a standard method is to set an action penalty in the reward to find the optimal motion considering the energy expenditure. This method is widely used for the simplicity of implementation. However, since the reward is a linear sum, if the penalty is too large, the system will fall into local minima and no moving solution can be obtained. In contrast, if the penalty is too small, the effect may not be sufficient. Therefore, it is necessary to adjust the amount of the penalty so that the agent always moves dynamically, and the energy-saving effect is sufficient. Nevertheless, since adjusting the hyperparameters is computationally expensive, we need a learning method that is robust to the penalty setting problem. We investigated on the Spiking Neural Network (SNN), which has been attracting attention for its computational efficiency and neuromorphic architecture. We conducted gait experiments using a hexapod agent while varying the energy penalty settings in the simulation environment. By applying SNN to the conventional state-of-the-art DRL algorithms, we examined whether the agent could explore for an optimal gait with a larger penalty variation and obtain an energy-efficient gait verified with Cost of Transport (CoT), a metric of energy efficiency for gait. Soft Actor-Critic (SAC)+SNN resulted in a CoT of 1.64, Twin Delayed Deep Deterministic policy gradient (TD3)+SNN resulted in a CoT of 2.21, and Deep Deterministic policy gradient (DDPG)+SNN resulted in a CoT of 2.08 (1.91 for normal SAC, 2.38 for TD3, and 2.40 for DDPG). DRL combined with SNN succeeded in learning more energy efficient gait with lower CoT.Katsumi NayaKyo KutsuzawaDai OwakiMitsuhiro HayashibeIEEEarticleSpiking neural networkdeep reinforcement learningenergy efficiencyhexapod gaitspatio-temporal backpropagationElectrical engineering. Electronics. Nuclear engineeringTK1-9971ENIEEE Access, Vol 9, Pp 150345-150354 (2021)
institution DOAJ
collection DOAJ
language EN
topic Spiking neural network
deep reinforcement learning
energy efficiency
hexapod gait
spatio-temporal backpropagation
Electrical engineering. Electronics. Nuclear engineering
TK1-9971
spellingShingle Spiking neural network
deep reinforcement learning
energy efficiency
hexapod gait
spatio-temporal backpropagation
Electrical engineering. Electronics. Nuclear engineering
TK1-9971
Katsumi Naya
Kyo Kutsuzawa
Dai Owaki
Mitsuhiro Hayashibe
Spiking Neural Network Discovers Energy-Efficient Hexapod Motion in Deep Reinforcement Learning
description In Deep Reinforcement Learning (DRL) for robotics application, it is important to find energy-efficient motions. For this purpose, a standard method is to set an action penalty in the reward to find the optimal motion considering the energy expenditure. This method is widely used for the simplicity of implementation. However, since the reward is a linear sum, if the penalty is too large, the system will fall into local minima and no moving solution can be obtained. In contrast, if the penalty is too small, the effect may not be sufficient. Therefore, it is necessary to adjust the amount of the penalty so that the agent always moves dynamically, and the energy-saving effect is sufficient. Nevertheless, since adjusting the hyperparameters is computationally expensive, we need a learning method that is robust to the penalty setting problem. We investigated on the Spiking Neural Network (SNN), which has been attracting attention for its computational efficiency and neuromorphic architecture. We conducted gait experiments using a hexapod agent while varying the energy penalty settings in the simulation environment. By applying SNN to the conventional state-of-the-art DRL algorithms, we examined whether the agent could explore for an optimal gait with a larger penalty variation and obtain an energy-efficient gait verified with Cost of Transport (CoT), a metric of energy efficiency for gait. Soft Actor-Critic (SAC)+SNN resulted in a CoT of 1.64, Twin Delayed Deep Deterministic policy gradient (TD3)+SNN resulted in a CoT of 2.21, and Deep Deterministic policy gradient (DDPG)+SNN resulted in a CoT of 2.08 (1.91 for normal SAC, 2.38 for TD3, and 2.40 for DDPG). DRL combined with SNN succeeded in learning more energy efficient gait with lower CoT.
format article
author Katsumi Naya
Kyo Kutsuzawa
Dai Owaki
Mitsuhiro Hayashibe
author_facet Katsumi Naya
Kyo Kutsuzawa
Dai Owaki
Mitsuhiro Hayashibe
author_sort Katsumi Naya
title Spiking Neural Network Discovers Energy-Efficient Hexapod Motion in Deep Reinforcement Learning
title_short Spiking Neural Network Discovers Energy-Efficient Hexapod Motion in Deep Reinforcement Learning
title_full Spiking Neural Network Discovers Energy-Efficient Hexapod Motion in Deep Reinforcement Learning
title_fullStr Spiking Neural Network Discovers Energy-Efficient Hexapod Motion in Deep Reinforcement Learning
title_full_unstemmed Spiking Neural Network Discovers Energy-Efficient Hexapod Motion in Deep Reinforcement Learning
title_sort spiking neural network discovers energy-efficient hexapod motion in deep reinforcement learning
publisher IEEE
publishDate 2021
url https://doaj.org/article/335681a7613746e5a4b027d670d343bf
work_keys_str_mv AT katsuminaya spikingneuralnetworkdiscoversenergyefficienthexapodmotionindeepreinforcementlearning
AT kyokutsuzawa spikingneuralnetworkdiscoversenergyefficienthexapodmotionindeepreinforcementlearning
AT daiowaki spikingneuralnetworkdiscoversenergyefficienthexapodmotionindeepreinforcementlearning
AT mitsuhirohayashibe spikingneuralnetworkdiscoversenergyefficienthexapodmotionindeepreinforcementlearning
_version_ 1718425215926534144