About This Project

Cutting and suturing soft tissues during surgery involves precise movements. Data driven estimation has shown good outcomes in predicting the properties of soft tissues, and deep reinforcement learning environments that use real-tie physical simulation can be very accurate with surgical tasks. Our goal is to apply data estimation to the da Vinci Surgical Robot to improve its performance on cutting and suturing tasks.

Ask the Scientists

What is the context of this research?

Learning grasp detection with human dexterity is a remaining challenge in the robot exploration world. Several methods such as deep learning, CNNs and Unets, Q-learning and other RL methods, have been used to efficiently accomplish this objective. Manipulating soft bodies face 4 main challenges: deformations strongly impact the contact surface and object dynamics. Most grasping strategies and patterns for rigid objects are different from the ones for deformable bodies. Their physics properties are not fully understand and cannot be simulated using simple algorithms, hence FEM have shown to be an interesting tool for this analysis. Therefore, this project will contribute to reach the frontier of increasing the level of autonomy of Da Vinci robot.

What is the significance of this project?

Autonomous robotic systems will reduce the surgeon’s burden in routine procedures, such as picking tissues, cutting and suturing in an autonomous setting or telemedicine. Images from the endoscopic camera are the primary source of information in surgical procedures, hence improving the characterizations of objects and tissues will improve the accuracy and precision on visualizing those. Therefore, this study can potentially simplify the effort perform by the surgeon in long and complex procedures and in settings where there are not enough nurses and personnel to assisted the doctor during the surgery. This will help the surgeon to prioritize his/her focus on the more critical tasks. Giving robot autonomy helps increasing the precision of many tasks that are performed in suregery.

What are the goals of the project?

Our main objectives are based on training the da Vinci surgical robot to perform cutting, tearing and suturing with deformable tissues. Hence, data driven estimation through a deep learning algorithm will be used to train the robot on these tasks in an autonomous way. Furthermore, finite element analysis that will conducted using SOFA (Simulation Open Framework Architecture) will predict the behavior of the tissue and compare this with real experiments to implement these in autonomous and telemedicine interventions. The primary metrics keys for this project will be the accuracy of the predictions of different points in the tissue compared with the real experiment and the computation time that has to match on online real procedure.