[1]高磊,胡陟.软组织牵拉中的自适应Transformer柔顺力控制[J].中国医学物理学杂志,2026,43(4):503-509.[doi:DOI:10.3969/j.issn.1005-202X.2026.04.013]
　GAO Lei,HU Zhi.Adaptive Transformer-based compliant force control for soft tissue retraction[J].Chinese Journal of Medical Physics,2026,43(4):503-509.[doi:DOI:10.3969/j.issn.1005-202X.2026.04.013]
点击复制

软组织牵拉中的自适应Transformer柔顺力控制()

《中国医学物理学杂志》[ISSN:1005-202X/CN:44-1351/R]

卷:
43卷
期数:
2026年第4期
页码:
503-509
栏目:
生物材料与力学
出版日期:
2026-04-28

文章信息/Info

Title:
Adaptive Transformer-based compliant force control for soft tissue retraction
文章编号:
1005-202X（2026）04-0503-07
作者:
高磊; 胡陟
上海工程技术大学电子电气工程学院， 上海 201620
Author(s):
GAO Lei;  HU Zhi
School of Electronic and Electrical Engineering, Shanghai University of Engineering Science, Shanghai 201620, China
关键词:
自适应阻抗控制; 力反馈; 预测控制; Transformer
Keywords:
Keywords: adaptive impedance control force feedback predictive control Transformer
分类号:
R318；TP242
DOI:
DOI:10.3969/j.issn.1005-202X.2026.04.013
文献标志码:
A
摘要:
针对复杂软组织环境中手术机器人面临的环境刚度突变与动态不确定性挑战，传统阻抗控制适应性不足问题，提出融合Transformer序列预测机制的自适应阻抗控制策略。通过对历史接触力与位移序列进行多变量建模，利用Transformer模型预测环境刚度、接触位置及控制参数演化趋势，实现控制参数前馈调节，增强系统对非线性、非平稳环境的适应能力。软组织牵拉实验中，与固定阻抗控制、传统自适应阻抗控制及模型预测控制（MPC）的对比显示：10 N目标力场景下，所提方法1.7 s达到目标力，稳态误差±0.2 N左右，最大力过冲约0.26 N；环境刚度突变时，恢复时间0.35 s，收敛时间为1.43 s；扰动实验中，正弦扰动条件下最大偏差3.17 N，积分绝对误差（IAE）降至3.52，阶跃扰动条件下恢复时间0.34 s。实验证实该方法在响应速度、稳态精度、过冲抑制及抗扰性能上均优于对比方法，为医疗机器人高精度柔顺控制提供了新的技术支撑。
Abstract:
Abstract: An adaptive impedance control strategy incorporating Transformer-based sequence prediction mechanisms is proposed to address the challenges of environmental stiffness variations and dynamic uncertainties faced by surgical robots in complex soft tissue environments, in which traditional impedance control methods struggle with insufficient adaptability. The proposed method performs multivariate modeling on historical contact force and displacement sequences, and utilizes Transformer models to predict the evolution trends of environmental stiffness, contact positions, and control parameters, thereby achieving feedforward adjustment of control parameters and improving system adaptability to nonlinear and non-stationary environments. Comparative experiments against fixed impedance control, conventional adaptive impedance control, and model predictive control during soft tissue retraction demonstrate that, under a 10 N target-force condition, the proposed method reaches the target force within 1.7 s, with a steady-state error of about ±0.2 N and a maximum force overshoot of about 0.26 N during environmental stiffness variations, the recovery time is 0.35 s and the settling time is 1.43 s in disturbance experiments, under sinusoidal disturbance, the maximum deviation is 3.17 N and the integral absolute error (IAE) is reduced to 3.52, while under step disturbance, the recovery time is 0.34 s. Experimental results demonstrate that the proposed method outperforms comparative approaches in response speed, steady-state accuracy, overshoot suppression, and disturbance rejection performance, providing new technical support for high-precision compliant control of medical robots.

相似文献/References:

[1]方程,胡陟.双臂手术机器人力控制优化[J].中国医学物理学杂志,2025,42(3):397.[doi:10.3969/j.issn.1005-202X.2025.03.017]
　FANG Cheng,HU Zhi.Force control optimization for dual-arm surgical robot[J].Chinese Journal of Medical Physics,2025,42(4):397.[doi:10.3969/j.issn.1005-202X.2025.03.017]

备注/Memo

备注/Memo:
【收稿日期】2025-12-16 【基金项目】国家自然科学基金（62003207）；国家重点研发计划（2019YFC0119303）；中国博士后基金面上资助项目（2021M690629） 【作者简介】高磊，硕士研究生，研究方向：机器人技术，E-mail: 2449300781@qq.com 【通信作者】胡陟，副教授，研究方向：力触觉反馈、力控制，E-mail: huzhi26@126.com
更新日期/Last Update: 2026-04-29