Data-driven control and grasping of multi-finger hybrid robotic arm
SUN Lang-lang
HE Shao-ying
XU Yun-wen
CHEN You-ren
PAN Xu-hua
Abstract：In this paper,a multi-finger hybrid robotic arm control system and its grasping control method are designed.The rigid-flexible hybrid structure enhances both the accuracy of control and the safety of object interaction during grasping tasks.Kinematic modeling of the rigid and flexible components of the robotic arm is performed using screw theory and the piecewise constant curvature method,respectively.An integrated model of the rigid-flexible hybrid robotic arm,based on the Jacobian matrix,is then derived.To mitigate the impact of model inaccuracies on system performance,a novel approach is proposed that combines data-driven techniques with model predictive control.This method replaces portions of the imprecise system model with historical data and constructs the current state and control inputs through a linear combination of this data.The effectiveness of this approach in controlling the posture of the flexible gripper is demonstrated through favorable trajectory tracking results in simulations.Building upon this,an accurate grasping and enveloping grasping method based on the posture of the grasped object is developed,and the validity of the method is confirmed through ArUco cube grasping experiments.
Keywords：hybrid robotic armmodel predictive controlinput mappingdata-drivengrasping control
Publication Date：2025-12-30
Online Publishing Date：2026-03-05(First online date of this platform, not the publication date of the document)
Pages：10( 2477-2486 )