.png){kind=icon}
.png){kind=icon}
Return
KoopFO: Koopman operator-based force observer for tendon-driven continuum robots
.png){kind=icon}
PRE
AI
L
L
Z
G
C
Z
DOI：10.1016/j.robot.2026.105446.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
Accurately estimating contact forces is essential for the safe and effective operation of continuum robots under external loading. However, significant challenges arise from their inherent compliance and nonlinear statics. This paper presents a Koopman operator-based force observer (KoopFO) synthesis method for varying-curvature continuum robots, which directly estimates end-effector forces from measured tip pose. Leveraging Koopman operator theory, the method approximates the robot's static behavior using arc length as the independent variable, yielding a globally linear representation of the nonlinear statics within a lifted state space. The Koopman operator is optimized via extended dynamic mode decomposition (EDMD) applied to both simulation and experimental data. A force observer is synthesized by integrating the approximated linear static model with a Kalman filter. This enables the iterative estimation of the external tip force and its associated uncertainty without requiring local linearization of the robot's underlying static model. The proposed method is validated through comprehensive simulations and experiments under diverse loading conditions. Results demonstrate high estimation accuracy, achieving a force relative error of 6.7%.
Keywords:
Continuum robots
Koopman operator
Kalman filter
Contact force detection
Journal
IF:
5.2
Papers:
621
Citations:
1.0W