MPC for Suturing Stitch Automation

Robot-assisted surgery (RAS) requires effective control strategies to ensure safety and accuracy while respecting the physical limits of the robot during tasks such as suturing and tissue manipulation. Model Predictive Control (MPC), with its inherent capability to handle complex dynamic systems, predict the future response and enforce constraints, is well-suited for these tasks. In this paper, MPC is employed to automate the suturing stitch task by mapping the operational space trajectory to the joint space while ensuring compliance with system kinematics constraints and safety requirements. To address varying requirements during suturing sub-tasks, two different objective functions and their corresponding constraint sets are used. The proposed framework is implemented using the ACADO toolkit to solve the Optimal Control Problem (OCP) and ROS to connect ACADO to CoppeliaSim/DVRK. Validation through simulations in CoppeliaSim and real-time experiments on the DVRK demonstrated that our approach achieved a positional/orientational accuracy of less than <inline-formula> <tex-math notation="LaTeX">$1mm/4 ^{\circ }$ </tex-math></inline-formula> in simulations, and an error norm of approximately <inline-formula> <tex-math notation="LaTeX">$1.9mm$ </tex-math></inline-formula> in real world implementations, confirming its effectiveness in automating suturing task.