An Adaptive Control Method and Learning Strategy for Ultrasound-Guided Puncture Robot
The development of a new generation of minimally invasive surgery is mainly reflected in robot-assisted diagnosis and treatment methods and their clinical applications. It is a clinical concern for robot-assisted surgery to use a multi-joint robotic arm performing human ultrasound scanning or ultras...
| Main Authors: | , , , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
MDPI AG
2024-01-01
|
| Series: | Electronics |
| Subjects: | |
| Online Access: | https://www.mdpi.com/2079-9292/13/3/580 |
| _version_ | 1797318897305649152 |
|---|---|
| author | Tao Li Quan Zeng Jinbiao Li Cheng Qian Hanmei Yu Jian Lu Yi Zhang Shoujun Zhou |
| author_facet | Tao Li Quan Zeng Jinbiao Li Cheng Qian Hanmei Yu Jian Lu Yi Zhang Shoujun Zhou |
| author_sort | Tao Li |
| collection | DOAJ |
| description | The development of a new generation of minimally invasive surgery is mainly reflected in robot-assisted diagnosis and treatment methods and their clinical applications. It is a clinical concern for robot-assisted surgery to use a multi-joint robotic arm performing human ultrasound scanning or ultrasound-guided percutaneous puncture. Among them, the motion control of the robotic arm, and the guiding and contact scanning processes of the ultrasonic (US-) probe determine the diagnosis effect, as well as the accuracy and safety of puncture surgery. To address these challenges, this study developed an intelligent robot-assisted system integrating autonomous US inspection and needle positioning, which has relation to several intelligent algorithms such as adaptive flexible control of the robot arm, autonomous US-scanning, and real-time attitude adjustment of the puncture needle. To improve the cooperativity of the spatial operation of the robot end-effector, we propose an adaptive flexible control algorithm that allows the operator to control the robot arm flexibly with low damping. To achieve the stability and uniformity of contact detection and imaging, we introduced a self-scanning method of US-probe based on reinforcement learning and built a software model of variable stiffness based on MuJoco to verify the constant force and velocity required by the end mechanism. We conducted a fixed trajectory scanning experiment at a scanning speed of 0.06 m/s. The force curve generally converges towards the desired contact force of 10 N, with minor oscillations around this value. For surgical process monitoring, we adopted the puncture needle detection algorithm based on Unet++ to acquire the position and attitude information of the puncture needle in real time. In short, we proposed and verified an adaptive control method and learning strategy by using an UR robotic arm equipped with a US-probe and puncture needle, and we improved the intelligence of the US-guided puncture robot. |
| first_indexed | 2024-03-08T03:59:03Z |
| format | Article |
| id | doaj.art-d489bde7c86e4441a4bed2e914763d3e |
| institution | Directory Open Access Journal |
| issn | 2079-9292 |
| language | English |
| last_indexed | 2024-03-08T03:59:03Z |
| publishDate | 2024-01-01 |
| publisher | MDPI AG |
| record_format | Article |
| series | Electronics |
| spelling | doaj.art-d489bde7c86e4441a4bed2e914763d3e2024-02-09T15:10:42ZengMDPI AGElectronics2079-92922024-01-0113358010.3390/electronics13030580An Adaptive Control Method and Learning Strategy for Ultrasound-Guided Puncture RobotTao Li0Quan Zeng1Jinbiao Li2Cheng Qian3Hanmei Yu4Jian Lu5Yi Zhang6Shoujun Zhou7Shenzhen Institutes of Advanced Technology, Chinese Academy of Sciences, Shenzhen 518055, ChinaShenzhen Institutes of Advanced Technology, Chinese Academy of Sciences, Shenzhen 518055, ChinaShenzhen Institutes of Advanced Technology, Chinese Academy of Sciences, Shenzhen 518055, ChinaShenzhen Institutes of Advanced Technology, Chinese Academy of Sciences, Shenzhen 518055, ChinaShenzhen Institutes of Advanced Technology, Chinese Academy of Sciences, Shenzhen 518055, ChinaCenter of Interventional Radiology & Vascular Surgery, Department of Radiology, Zhongda Hospital, Southeast University, 87 Dingjiaqiao Road, Nanjing 210009, ChinaCenter of Interventional Radiology & Vascular Surgery, Department of Radiology, Zhongda Hospital, Southeast University, 87 Dingjiaqiao Road, Nanjing 210009, ChinaShenzhen Institutes of Advanced Technology, Chinese Academy of Sciences, Shenzhen 518055, ChinaThe development of a new generation of minimally invasive surgery is mainly reflected in robot-assisted diagnosis and treatment methods and their clinical applications. It is a clinical concern for robot-assisted surgery to use a multi-joint robotic arm performing human ultrasound scanning or ultrasound-guided percutaneous puncture. Among them, the motion control of the robotic arm, and the guiding and contact scanning processes of the ultrasonic (US-) probe determine the diagnosis effect, as well as the accuracy and safety of puncture surgery. To address these challenges, this study developed an intelligent robot-assisted system integrating autonomous US inspection and needle positioning, which has relation to several intelligent algorithms such as adaptive flexible control of the robot arm, autonomous US-scanning, and real-time attitude adjustment of the puncture needle. To improve the cooperativity of the spatial operation of the robot end-effector, we propose an adaptive flexible control algorithm that allows the operator to control the robot arm flexibly with low damping. To achieve the stability and uniformity of contact detection and imaging, we introduced a self-scanning method of US-probe based on reinforcement learning and built a software model of variable stiffness based on MuJoco to verify the constant force and velocity required by the end mechanism. We conducted a fixed trajectory scanning experiment at a scanning speed of 0.06 m/s. The force curve generally converges towards the desired contact force of 10 N, with minor oscillations around this value. For surgical process monitoring, we adopted the puncture needle detection algorithm based on Unet++ to acquire the position and attitude information of the puncture needle in real time. In short, we proposed and verified an adaptive control method and learning strategy by using an UR robotic arm equipped with a US-probe and puncture needle, and we improved the intelligence of the US-guided puncture robot.https://www.mdpi.com/2079-9292/13/3/580robot-assisted systemultrasound scanningflexible controlreinforcement learningPPO |
| spellingShingle | Tao Li Quan Zeng Jinbiao Li Cheng Qian Hanmei Yu Jian Lu Yi Zhang Shoujun Zhou An Adaptive Control Method and Learning Strategy for Ultrasound-Guided Puncture Robot Electronics robot-assisted system ultrasound scanning flexible control reinforcement learning PPO |
| title | An Adaptive Control Method and Learning Strategy for Ultrasound-Guided Puncture Robot |
| title_full | An Adaptive Control Method and Learning Strategy for Ultrasound-Guided Puncture Robot |
| title_fullStr | An Adaptive Control Method and Learning Strategy for Ultrasound-Guided Puncture Robot |
| title_full_unstemmed | An Adaptive Control Method and Learning Strategy for Ultrasound-Guided Puncture Robot |
| title_short | An Adaptive Control Method and Learning Strategy for Ultrasound-Guided Puncture Robot |
| title_sort | adaptive control method and learning strategy for ultrasound guided puncture robot |
| topic | robot-assisted system ultrasound scanning flexible control reinforcement learning PPO |
| url | https://www.mdpi.com/2079-9292/13/3/580 |
| work_keys_str_mv | AT taoli anadaptivecontrolmethodandlearningstrategyforultrasoundguidedpuncturerobot AT quanzeng anadaptivecontrolmethodandlearningstrategyforultrasoundguidedpuncturerobot AT jinbiaoli anadaptivecontrolmethodandlearningstrategyforultrasoundguidedpuncturerobot AT chengqian anadaptivecontrolmethodandlearningstrategyforultrasoundguidedpuncturerobot AT hanmeiyu anadaptivecontrolmethodandlearningstrategyforultrasoundguidedpuncturerobot AT jianlu anadaptivecontrolmethodandlearningstrategyforultrasoundguidedpuncturerobot AT yizhang anadaptivecontrolmethodandlearningstrategyforultrasoundguidedpuncturerobot AT shoujunzhou anadaptivecontrolmethodandlearningstrategyforultrasoundguidedpuncturerobot AT taoli adaptivecontrolmethodandlearningstrategyforultrasoundguidedpuncturerobot AT quanzeng adaptivecontrolmethodandlearningstrategyforultrasoundguidedpuncturerobot AT jinbiaoli adaptivecontrolmethodandlearningstrategyforultrasoundguidedpuncturerobot AT chengqian adaptivecontrolmethodandlearningstrategyforultrasoundguidedpuncturerobot AT hanmeiyu adaptivecontrolmethodandlearningstrategyforultrasoundguidedpuncturerobot AT jianlu adaptivecontrolmethodandlearningstrategyforultrasoundguidedpuncturerobot AT yizhang adaptivecontrolmethodandlearningstrategyforultrasoundguidedpuncturerobot AT shoujunzhou adaptivecontrolmethodandlearningstrategyforultrasoundguidedpuncturerobot |