Research on Robotic Arm Motion Path Based on Reinforcement Learning and LLM

Shihan Zhe

doi:10.54254/2755-2721/2025.LD28451

Applied and Computational EngineeringOpen access

Research on Robotic Arm Motion Path Based on Reinforcement Learning and LLM

Research Article

Open Access

Research on Robotic Arm Motion Path Based on Reinforcement Learning and LLM

Shihan Zhe ^1*

¹ Sino-German College of Applied Sciences, Tongji University, Shanghai, 201804, China

^*Corresponding author: shihan.zhe@outlook.com

Published on 22 October 2025

ACE Vol.196

ISSN (Print): 2755-273X

ISSN (Online): 2755-2721

ISBN (Print): 978-1-80590-451-9

ISBN (Online): 978-1-80590-452-6

Download Cover

Abstract

Robot control is a current research hotspot, with robotic arm trajectory planning being a key direction. However, traditional methods exhibit insufficient adaptability and flexibility, making it difficult to meet the demands of complex tasks and dynamic environments. This paper proposes a path planning system based on the collaboration of reinforcement learning (RL) and large language models (LLM). The system consists of three modules: environment perception and LLM-based task decomposition and scheduling, RL-based trajectory planning, and motion command generation. By integrating the cognitive capabilities of LLM with the optimization capabilities of RL, the system enables task-driven robotic arm trajectory planning. In terms of design, the upper layer employs LLM for task analysis and high-level command generation, while the lower layer uses RL for trajectory optimization, forming a hierarchical collaborative mechanism. To verify its effectiveness, experiments were conducted on both simulated and real COBOT platforms for a static block-grabbing task, comparing three schemes: pure RL, pure LLM, and the proposed LLM-RL fusion. Results show that the LLM-RL approach outperforms the baselines in terms of average path length and execution time, while also significantly improving RL training efficiency.

Keywords:

robotic arm, reinforcement learning, large language model, path planning

View PDF

References

[1]. Zong, C., Lu, L., Lei, X., & Zhao, P. (2017). A path planning approach for multi-dof spatial manipulator via A* algorithm. Journal of Hefei University of Technology (Natural Science), 40(2), 164-168.

[2]. Yaqiu, L., Hanchen, Z., Xun, L., & Yan, X. (2021). An Improved RRT Based Obstacle Avoidance Path Planning Algorithm for Industrial Robot. Information and Control, 50(2), 235-246.

[3]. Zou, Y., Li, L., & Gao, Z. (2019). Obstacle avoidance path planning for harvesting robot arm based on improved PRM. Transducer and Microsystem Technologies, 38(1), 52-56.

[4]. Li, Y., & Xu, D. (2020). Cooperative path planning of dual-arm robot based on attractive force self-adaptive step size RRT. Robot, 42(5), 606-616.

[5]. Wang, Y., Liu, Y., Jia, H., & Xue, G. (2022). Path planning of mechanical arm based on intensified RRT algorithm. J. Shandong Univ.(Eng. Sci.), 52, 123-130.

[6]. Ma, R., Luijkx, J., Ajanovic, Z., & Kober, J. (2024). Explorllm: Guiding exploration in reinforcement learning with large language models. arXiv preprint arXiv: 2403.09583.

[7]. Kheirandish, A., Xu, D., & Fekri, F. (2024). LLM-Augmented Symbolic Reinforcement Learning with Landmark-Based Task Decomposition. arXiv preprint arXiv: 2410.01929.

[8]. Shukla, Y., Gao, W., Sarathy, V., Velasquez, A., Wright, R., & Sinapov, J. (2023). Lgts: Dynamic task sampling using llm-generated sub-goals for reinforcement learning agents. arXiv preprint arXiv: 2310.09454.

[9]. Shek, C. L., & Tokekar, P. (2025). Option Discovery Using LLM-guided Semantic Hierarchical Reinforcement Learning. arXiv preprint arXiv: 2503.19007.

[10]. Siedler, P. D., & Gemp, I. (2025). LLM-Mediated Guidance of MARL Systems. arXiv preprint arXiv: 2503.13553.