Background The tendon‐sheath‐system (TSS) is an excellent medium for remote power transmission, which is widely used in laparoscopic surgery robots. Since the operation process requires the robot to move continuously, this time‐varying characteristic further aggravates the force and position transmission loss caused by the nonlinear friction of TSS, which affects the control accuracy of the surgical robot. Method A time‐varying tendon‐sheath transmission model (RT model) is proposed. A feedforward control system is designed to improve tendon‐sheath transmission accuracy. Furthermore, a tendon‐sheath transmission model with velocity characteristics (RV model) is established. Result Force, position, and velocity experiments were carried out on the platform of TSS with a robotic arm. The results show that the R‐square values of force and position compensation are at least 96.57% and 99.16%. Conclusion The proposed RT and RV models are effective in compensating for the TSS transmission loss during the operation of the surgical robot.

The inaccurate force and position control of tendon sheath system (TSS) due to nonlinear friction during surgery seriously hinders its development in the field of precision surgical robots. To this end, this paper proposes a time-varying bending angle estimation method under the state of sensorless offline identification combined with robot kinematics by analyzing the friction of the TSS and the deformation of the robot during the movement, and establishes a force and position transfer model with time-varying path trajectory (SJM model). The model uses B-spline curve to fit tendon-sheath trajectory. In order to further improve the control accuracy of force and position, a new intelligent feedforward control strategy that integrates the SJM model and a neural network algorithm is proposed. In order to gain an in-depth understanding of the transmission process of force and position and to demonstrate the validity of the SJM model, an experimental platform for the TSS was built. A feedforward control system under the MATLAB environment was built with the aim of verifying the accuracy of the intelligent feedforward control strategy. The system innovatively combines the SJM model with BP and RBF neural networks, respectively. The experimental results showed that the correlation coefficients (R2) of force and position transfer are above 99.10% and 99.48%, respectively. Ultimately, we compared the intelligent feedforward and intelligent control strategy under a single neural network, and observed that the intelligent feedforward control strategy has a better effect.