Background: Robotic camera holders in laparoscopic surgery improve surgical efficiency and reduce the burden on medical specialists.
Methods: We propose a multi-task compliant control framework that integrates deep learning methods with robot kinematics. This framework addresses key challenges in surgical procedures, such as maintaining the remote center of motion (RCM) constraint and achieving autonomous field of view (FOV) adjustment.
Results: Experimental results demonstrate that our framework follows various trajectories with mean response time of less than 2 s, maximum RCM constraint error of less than 5 mm, mean tracking error of less than 20 pixels, and mean depth error of less than 2.5 mm. Additionally, its scalability enabled successful integration of a virtual fixture to prevent tissue collisions.
Conclusion: Our framework enables autonomous, rapid, and safe laparoscope manipulation, enhancing the continuity and efficiency of surgical procedures while conserving specialist healthcare resources.
Keywords: deep learning methods; laparoscopy; medical robots and systems; robot‐assisted surgery.
© 2025 John Wiley & Sons Ltd.