Vision-Based Contact Force Sensing in Robotic Surgery: A Technical Review

摘要

Contact force sensing plays a fundamental role in enabling precise and safe manipulations in surgical robotics. While traditional force sensing methods predominantly rely on physical sensors, they encounter significant limitations including integration complexity, limited durability, and scalability constraints in surgical environments. This paper presents a comprehensive survey of vision-based force sensing approaches that leverage computer vision and image processing techniques to infer contact forces without dedicated physical sensors. The survey systematically analyzes state-of-the-art methodologies encompassing vision-based force estimation, surface deformation analysis, and machine learning-driven perception frameworks. The underlying principles, algorithmic architectures, and surgical applications of these approaches are examined in detail. Critical challenges including real-time processing requirements, robustness to environmental variations, and cross-task generalizability are thoroughly investigated. The survey highlights the distinctive advantages of purely vision-based approaches, including reduced hardware complexity, enhanced adaptability, and seamless integration potential with existing surgical robotic platforms. Furthermore, this work identifies critical open research questions and promising future directions for advancing vision-based force sensing technologies in next-generation surgical robotics. This comprehensive review serves as an authoritative reference for researchers and practitioners working to overcome the limitations of traditional contact force sensing through vision-based solutions in surgical applications.