Fuzzy C-Regression Clustering Algorithm Based 7-Dof Redundant Manipulators Inverse Dynamics Control

High-precision tracking control of redundant manipulators is essential in robotic applications, but it often requires an exact dynamic model. However, the complex mechanical structure of these manipulators makes it challenging to apply traditional model-based control methods effectively. To address...

Celý popis

Uloženo v:
Podrobná bibliografie
Vydáno v:Chinese Control and Decision Conference s. 319 - 324
Hlavní autoři: Zhao, Jiayu, Zhao, Tao, Yang, Hainan, Pan, Jianhui
Médium: Konferenční příspěvek
Jazyk:angličtina
Vydáno: IEEE 16.05.2025
Témata:
Clustering algorithms
Fuzzy C-regression model
Fuzzy sets
Heuristic algorithms
hyper-plane-shaped clustering
interval type-2 T-S fuzzy sets
inverse dynamics control
Manipulator dynamics
Real-time systems
redundant manipulators
Robustness
Simulation
Torque
Trajectory
Uncertainty
ISSN:1948-9447
On-line přístup:Získat plný text
Tagy: Přidat tag
Žádné tagy, Buďte první, kdo vytvoří štítek k tomuto záznamu!
Popis
Shrnutí:High-precision tracking control of redundant manipulators is essential in robotic applications, but it often requires an exact dynamic model. However, the complex mechanical structure of these manipulators makes it challenging to apply traditional model-based control methods effectively. To address this challenge, we propose a data-driven approach using the interval type-2 (IT2) fuzzy c-regression clustering (FCRM) algorithm to develop an inverse dynamics model for redundant manipulators. Utilizing the FCRM algorithm, the nonlinear inverse dynamics input-output data of the robotic arm is subjected to hyper-planeshaped (HPS) clustering, which results in the generation of a T-S submodel and its associated membership function (MF). With the aid of the IT2 fuzzy set, the T-S model captures unmodeled dynamics and uncertainty information of redundant manipulators. Offline data is utilized to learn the inverse dynamics TS submodel and its membership function, enabling real-time control of the manipulator input based on predicted control torque derived from the target trajectory. Simulation results on the Franka Emika Panda (7-DOF redundant manipulator) validate the feasibility of the proposed method.
ISSN:1948-9447
DOI:10.1109/CCDC65474.2025.11090180