Integration of an actor-critic model and generative adversarial networks for a Chinese calligraphy robot

Standard

Integration of an actor-critic model and generative adversarial networks for a Chinese calligraphy robot. / Wu, Ruiqi; Zhou, Changle ; Chao, Fei; Yang, Longzhi ; Lin, Chih-Min; Shang, Changjing.

In: Neurocomputing, Vol. 388, 07.05.2020, p. 12-23.

Research output: Contribution to journalArticlepeer-review

Author

Wu, Ruiqi ; Zhou, Changle ; Chao, Fei ; Yang, Longzhi ; Lin, Chih-Min ; Shang, Changjing. / Integration of an actor-critic model and generative adversarial networks for a Chinese calligraphy robot. In: Neurocomputing. 2020 ; Vol. 388. pp. 12-23.

Bibtex - Download

@article{ab2d0897726646fbb8c14a707956345d,
title = "Integration of an actor-critic model and generative adversarial networks for a Chinese calligraphy robot",
abstract = "As a combination of robotic motion planning and Chinese calligraphy culture, robotic calligraphy plays a significant role in the inheritance and education of Chinese calligraphy culture. Most existing calligraphy robots focus on enabling the robots to learn writing through human participation, such as human–robot interactions and manually designed evaluation functions. However, because of the subjectivity of art aesthetics, these existing methods require a large amount of implementation work from human engineers. In addition, the written results cannot be accurately evaluated. To overcome these limitations, in this paper, we propose a robotic calligraphy model that combines a generative adversarial network (GAN) and deep reinforcement learning to enable a calligraphy robot to learn to write Chinese character strokes directly from images captured from Chinese calligraphic textbooks. In our proposed model, to automatically establish an aesthetic evaluation system for Chinese calligraphy, a GAN is first trained to understand and reconstruct stroke images. Then, the discriminator network is independently extracted from the trained GAN and embedded into a variant of the reinforcement learning method, the “actor-critic model”, as a reward function. Thus, a calligraphy robot adopts the improved actor-critic model to learn to write multiple character strokes. The experimental results demonstrate that the proposed model successfully allows a calligraphy robot to write Chinese character strokes based on input stroke images. The performance of our model, compared with the state-of-the-art deep reinforcement learning method, shows the efficacy of the combination approach. In addition, the key technology in this work shows promise as a solution for robotic autonomous assembly.",
author = "Ruiqi Wu and Changle Zhou and Fei Chao and Longzhi Yang and Chih-Min Lin and Changjing Shang",
year = "2020",
month = may,
day = "7",
doi = "10.1016/j.neucom.2020.01.043",
language = "English",
volume = "388",
pages = "12--23",
journal = "Neurocomputing",
issn = "0925-2312",
publisher = "Elsevier",

}

RIS (suitable for import to EndNote) - Download

TY - JOUR

T1 - Integration of an actor-critic model and generative adversarial networks for a Chinese calligraphy robot

AU - Wu, Ruiqi

AU - Zhou, Changle

AU - Chao, Fei

AU - Yang, Longzhi

AU - Lin, Chih-Min

AU - Shang, Changjing

PY - 2020/5/7

Y1 - 2020/5/7

N2 - As a combination of robotic motion planning and Chinese calligraphy culture, robotic calligraphy plays a significant role in the inheritance and education of Chinese calligraphy culture. Most existing calligraphy robots focus on enabling the robots to learn writing through human participation, such as human–robot interactions and manually designed evaluation functions. However, because of the subjectivity of art aesthetics, these existing methods require a large amount of implementation work from human engineers. In addition, the written results cannot be accurately evaluated. To overcome these limitations, in this paper, we propose a robotic calligraphy model that combines a generative adversarial network (GAN) and deep reinforcement learning to enable a calligraphy robot to learn to write Chinese character strokes directly from images captured from Chinese calligraphic textbooks. In our proposed model, to automatically establish an aesthetic evaluation system for Chinese calligraphy, a GAN is first trained to understand and reconstruct stroke images. Then, the discriminator network is independently extracted from the trained GAN and embedded into a variant of the reinforcement learning method, the “actor-critic model”, as a reward function. Thus, a calligraphy robot adopts the improved actor-critic model to learn to write multiple character strokes. The experimental results demonstrate that the proposed model successfully allows a calligraphy robot to write Chinese character strokes based on input stroke images. The performance of our model, compared with the state-of-the-art deep reinforcement learning method, shows the efficacy of the combination approach. In addition, the key technology in this work shows promise as a solution for robotic autonomous assembly.

AB - As a combination of robotic motion planning and Chinese calligraphy culture, robotic calligraphy plays a significant role in the inheritance and education of Chinese calligraphy culture. Most existing calligraphy robots focus on enabling the robots to learn writing through human participation, such as human–robot interactions and manually designed evaluation functions. However, because of the subjectivity of art aesthetics, these existing methods require a large amount of implementation work from human engineers. In addition, the written results cannot be accurately evaluated. To overcome these limitations, in this paper, we propose a robotic calligraphy model that combines a generative adversarial network (GAN) and deep reinforcement learning to enable a calligraphy robot to learn to write Chinese character strokes directly from images captured from Chinese calligraphic textbooks. In our proposed model, to automatically establish an aesthetic evaluation system for Chinese calligraphy, a GAN is first trained to understand and reconstruct stroke images. Then, the discriminator network is independently extracted from the trained GAN and embedded into a variant of the reinforcement learning method, the “actor-critic model”, as a reward function. Thus, a calligraphy robot adopts the improved actor-critic model to learn to write multiple character strokes. The experimental results demonstrate that the proposed model successfully allows a calligraphy robot to write Chinese character strokes based on input stroke images. The performance of our model, compared with the state-of-the-art deep reinforcement learning method, shows the efficacy of the combination approach. In addition, the key technology in this work shows promise as a solution for robotic autonomous assembly.

U2 - 10.1016/j.neucom.2020.01.043

DO - 10.1016/j.neucom.2020.01.043

M3 - Article

VL - 388

SP - 12

EP - 23

JO - Neurocomputing

JF - Neurocomputing

SN - 0925-2312

ER -

Show download statistics
View graph of relations
Citation formats