Machine Learning Assisting Robots

Research in machine learning (ML)-assisted surgical robotics aims to improve camera control and autonomous navigation. Wagner et al. developed a laparoscopic surgery camera system that learned from experience, showing improved execution time and guidance quality. Kim et al. focus on autonomously inserting a needle into an eye model, proving safer and more precise than manual methods. The challenge lies in translating these autonomous systems into realistic surgical scenarios. Saeidi et al. make strides by automating small bowel anastomosis in an in vivo setting, outperforming manual and robot-assisted methods. Human–machine cooperation is explored, illustrating realistic implementations of automated subtasks within the surgical workflow. Meanwhile, the significance of reinforcement learning (RL) in surgical robotics is growing, outlining its challenges and potential applications. RL is explored in various subtasks such as knot-tying, suturing, and tissue manipulation. Despite current limitations, studies anticipate increasing success in applying RL to robotic surgical scenarios.