Autonomy is the next-level technology to be explored in the field of surgical robotics. We have developed automated intra-operative surgical procedures by leveraging newly designed surgical robotic prototypes with AI-enabled technologies for online analysis of surgical image input. The main impact of our work lies on the fact that we have currently pushed robot autonomy in clinical applicability the farthest, including the cadaver study for automatic endoscope control. Our autonomous planning and control framework also allows unified planning and control of various tasks (including debridgement, cutting, suturing, etc.) with quality comparable to novice surgeons. As one of the pioneer teams worldwide, we expect surgery autonomy to achieve significant workload relief to surgeons with more consistent outcomes in the near future.
Image-guided Automatic Robotic Surgery
Researchers
Introduction
The Main Impact
1
Surgical Visual Perception and Intelligence
- New dataset and algorithms for surgical workflow recognition, surgical instrument detection and segmentation, and needle pose estimation.
- Novel algorithms for 3D reconstruction of tissue surfaces from surgical videos.
2
Robot-Assisted Surgical System
- An open-source reinforcement learning-centered simulation platform, SurRoL, for surgical robot learning, compatible with the da Vinci Research Kit (dVRK).
- A one-surgeon-four-arm human-robot collaboration system for total laparoscopic hysterectomy, verified by cadaver study in Prince of Wales Hospital.
3
Image-guided Planning and Control
- A unified surgical task autonomy framework that automates various simulated surgical
tasks, achieving greater precision and consistency than novices. - A robotised colonoscope equipped with an FBG sensor enables automatic in-vivo
navigation using integrated shape and vision-based control. - A robust marker-less feature-based controller for soft object manipulation provides
deformation control, effectively handling various types of soft materials.
