Semi-Autonomous Laparoscopic Robot Docking With Learned Hand-Eye Information Fusion

Area of Science:

• Robotics
• Surgical Technology
• Computer Vision

Background:

• Key-hole docking in minimally invasive surgery requires high precision and safety.
• Current methods face challenges with accuracy and force compliance.
• Shared-control systems offer potential for enhanced surgical performance.

Purpose of the Study:

• To introduce a novel shared-control system for key-hole docking operations.
• To enhance docking precision and force-compliance safety.
• To improve pose estimation accuracy in complex surgical environments.

Main Methods:

• Developed a shared-control system integrating a commercial camera with occlusion-robust pose estimation.
• Implemented a hand-eye information fusion technique for improved accuracy.
• Generated a self-supervised dataset for training a hand-eye information fusion network model.

Main Results:

• Pose estimation accuracy improved compared to traditional methods (observation-only, hand-eye calibration, state estimation filters).
• Reduced position dispersion (1.23±0.81 mm vs. 2.47±1.22 mm) in phantom experiments.
• Reduced force dispersion (0.78±0.57 N vs. 1.15±0.97 N) compared to the control group.

Conclusions:

• The novel system offers an anti-interference, steady, and precise solution for key-hole docking.
• Advancements in semi-autonomy co-manipulation enhance interaction and stability in surgical procedures.
• Potential applications extend beyond laparoscopic surgery to other minimally invasive procedures.