A Robotic Disassembly Planning Method for Retired Batteries Based on a Long Short-Term Memory Collaborative Framework
Symmetry · 05.06.2026 · 18 (6) · 981
DOI: 10.3390/sym18060981 ↗Kurz gefasst
Fachartikel in der Zeitschrift Symmetry, online erschienen am 05.06.2026. Verfasst von Jie Li und 3 weiteren Autor:innen. Die Arbeit nennt 22 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: Fundamental Research Funds for the Central Universities.
Abstract (Originalsprache)
This paper addresses non-steady-state scenarios in the human–robot collaborative disassembly process of retired power batteries, including component aging, ambiguous instructions, and sensor drift. In such scenarios, the robot exhibits execution robustness problems. This paper proposes a Planning Domain Definition Language (PDDL) generation framework that integrates long-term and short-term memory. The framework combines large language models with knowledge graphs as a long-term memory module for symbolic task decomposition and domain semantic rule generalization, while using meta-heuristic optimization algorithms as a short-term memory module to adapt and optimize action parameters based on real-time sensor feedback. Through this closed-loop mechanism that combines long-term memory guidance with short-term memory adaptation, the system addresses the limitation of traditional PDDL, which, when facing open, time-varying, and heterogeneous industrial disassembly scenarios, has symbolic action models that have difficulty capturing the uncertainty and unpredictable disturbances in real physical systems, limiting its practicality in complex non-steady-state scenarios. Furthermore, the system establishes a feedback mechanism from short-term memory to long-term memory, enhancing disassembly capabilities in non-steady-state environments by transforming scenario information into supplementary understanding. The research validates this method on a real disassembly platform. Compared with baselines of traditional PDDL, a planning method using only large language models (LLMs), and heuristic algorithms, this method achieved an 88.0% task success rate (significantly superior to the 38.0% of traditional PDDL).
Förderung
- Fundamental Research Funds for the Central Universities
Zitieren
APA
Li, J., Zhang, S., Si, J., & Bao, J. (2026). A Robotic Disassembly Planning Method for Retired Batteries Based on a Long Short-Term Memory Collaborative Framework. Symmetry, 18(6), 981. https://doi.org/10.3390/sym18060981
Vancouver
Li J, Zhang S, Si J, Bao J. A Robotic Disassembly Planning Method for Retired Batteries Based on a Long Short-Term Memory Collaborative Framework. Symmetry. 2026;18(6):981. doi:10.3390/sym18060981
BibTeX
@article{Li2026,
author = {Li, Jie and Zhang, Shuo and Si, Jiahui and Bao, Jinsong},
title = {A Robotic Disassembly Planning Method for Retired Batteries Based on a Long Short-Term Memory Collaborative Framework},
journal = {Symmetry},
year = {2026},
volume = {18},
number = {6},
pages = {981},
doi = {10.3390/sym18060981}
}
EUVIMED