Fuzzy Systems and its Applications

Fuzzy Systems and its Applications

Performing Complex Tasks by Robots Through Breaking Them Down into Simple Skills and Planning Using Large Language Models

Document Type : Original Article

Authors
Vali Derhami 1
Ehsan khademi 2
ALa Soltanzade 2
1 Yazd university
2 Yazd University
10.22034/jfsa.2025.528449.1278
Abstract
،This paper presents, an approach for breaking down complex robotic tasks into basic skills and combining them to accomplish complex tasks. Employing robots for complex tasks poses challenges such as time-consuming learning, lack of generalizability, and limited human interaction. This paper first introduces the idea of decomposing complex tasks into simple skills, enabling a robot to learn foundational skills and thereby perform a wide variety of complex tasks. This approach reduces learning time and achieves relatively good generalizability. For task planning, large language models (LLMs) such as ChatGPT and DeepSeek are used, and their capability to generate actionable plans by combining basic skills is evaluated. Two types of instructions are given to the system: one with clear inputs and the other with fuzzy values. The results demonstrate that the mentioned LLMs not only perform well with clear inputs but also exhibit an appropriate ability to comprehend task instructions and plan actions for the robot even when faced with fuzzy inputs. Thus, with their assistance, an interactive robotic system can be developed
Keywords
Large language models
Task decomposition
Robot
Planning
fuzzy logic
Subjects
 
[1]    J. Krüger, G. Schreck, and D. Surdilovic, ”Dual-arm robot for flexible and cooperative assembly,” CIRP Annals, vol. 60, no. 1, pp. 5–8, 2011.
[2]    P. Kazanzides et al., ”An open-source research kit for the da Vinci® Surgical System,” IEEE Robotics and Automation Magazine, vol. 24, no. 2, pp. 48–56, 2017.
 
[3]    M. Bac et al., ”Harvesting robots for high-value crops: State-of-the-art review and challenges ahead,” Journal of Field Robotics, vol. 37, no. 8, pp. 1310–1332, 2020.
[4]    R. Yamamoto et al., ”Service robot integration into daily human environments,” Robotics and Autonomous Systems, vol. 95, pp. 3–17, 2017.
[5]    A. Billard and D. Kragic, ”Trends and challenges in robot manipulation,” Science, vol. 364, no. 6446, p. eaat8414, 2019.
[6]    D. Whitney, M. Oberlin, and S. Tellex, ”Interpreting multimodal referring expressions in real time,” Robotics: Science and Systems (RSS), 2016.
[7]    J. Andreas, D. Klein, and S. Levine, ”Learning with latent language,” Proceedings of EMNLP, pp. 2166–2178, 2018.
[8]    T. Tellex et al., ”Understanding natural language commands for robotic navigation and mobile manipulation,” Journal of Experimental & Theoretical Artificial Intelligence, vol. 25, no. 4, pp. 409–434, 2013.
[9]    J. Xu, J. White, and M. Gombolay, ”Learning to compose skills for real-world robotic manipulation,” Proceedings of RSS, 2022.
[10]    Y. Ahn et al., ”Do as I can, not as I say: Grounding language in robotic affordances,” Proceedings of ICML, 2022.
[11]    B. Ichter et al., ”Language-model-based imitation learning for robotic manipulation,” arXiv preprint arXiv:2209.07753, 2022.
[12]    M. Jain et al., ”Bridging language and motor control with skill libraries,” Conference on Robot Learning (CoRL), 2023.
[13]    P. Kormushev, D. N. Nenchev, S. Calinon, and D. G. Caldwell, “Upper-body kinesthetic teaching of a free-standing humanoid robot,” Autonomous Robots, vol. 36, no. 1, pp. 123–143, 2013.
[14]    J. J. Craig, Introduction to Robotics: Mechanics and Control, 3rd ed., Pearson Prentice Hall, 2005.
[15]    A. Vaswani et al., “Attention is all you need,” NeurIPS, vol. 30, 2017.
[16]    T. Brown et al., “Language models are few-shot learners,” NeurIPS, vol. 33, pp. 1877–1901, 2020.
[17]    OpenAI, “GPT-4 Technical Report,” arXiv preprint arXiv:2303.08774, 2023.
[18]    Y. Kojima, S. Zhi, Y. Bai, and A. Zeng, “Large language models as zero-shot planners for robots,”
RSS, 2023.
Fuzzy Systems and its Applications
Volume 8, Issue 2 - Serial Number 17
Open Access Statement
December 2025
Pages 21-41

  • Receive Date 08 June 2025
  • Revise Date 30 September 2025
  • Accept Date 19 October 2025