Surgical robots, such as Intuitive Surgical’s da Vinci Surgical System, have brought about more efficient surgeries by improving dexterity and reducing surgeon fatigue through teleoperational control. While these systems are already providing great care to patients, they have also opened the door to a variety of research, including surgical task automation. Surgical task automation has furthermore been an increasing area of research in an effort to improve patient throughput, reduce quality-of-care variance among surgeries, and potentially deliver automated surgery in the future. We are developing algorithms and control policies that automate surgical tasks to work towards this future.

Learning to Automate Surgery

Reinforcement Learning (RL) is a machine learning framework for AI systems to solve complex problems. In recent years, success in solving challenging games and robotic manipulation tasks has increased, partly due to collaborative efforts on open-sourced environment simulators like OpenAI's Gym. We present the first open-sourced reinforcement learning environment for surgical robotics, called dVRL, which is functionally equivalent to Gym. dVRL enables prototyping and implementing state-of-art RL algorithms on surgical robotics problems, aiming to introduce autonomous robotic precision and accuracy during surgery. Combining dVRL with the da Vinci Surgical Research Kits network, we enable the surgical robotics community to leverage the newest RL strategies and RL scientists to develop algorithms for autonomous surgery challenges.

Robust Surgical Automation

Robustness is a critical factor in surgical automation since the nature of surgical procedures demands unwavering reliability and resilience. Robust surgical automation should be capable of seamlessly adapting to dynamic surgical scenarios, handling variability in patient anatomy, and quickly recovering from disruptions, minimizing the potential for adverse impacts on the patient's well-being. Achieving robustness in surgical automation is crucial for instilling confidence in both surgeons and patients, as it guarantees consistent and reliable performance, even in challenging or unpredictable circumstances.

Recent efforts in our group such as SURESTEP (Surgical Uncertainty-aware Robust ESTimation TrajEctory Protocol) present frameworks for uncertainty-aware trajectory optimization that enables robust automation by enhancing visual tool tracking accuracy. It optimizes trajectories from any policies to be robust to motion and observation uncertainties commonly encountered in surgical settings.