Research topic: Application to Needle Insertion
Scientifique Responsable: Hadrien Courtecuisse
We aim at developing an autonomous robotic system for RFA of deep liver tumors. During needle insertion, both the needle and the tissue undergo deformations and the targets, trajectories and obstacles may move simultaneously. Our approach is to develop control models allowing to correct and predict the deformation of structures (needle, tissues or the robot itself) and adapt the behavior of the robot in order to reach an objective. The key scientific problem we address in this PhD is to provide alternative solutions to control a robot based on inverse Finite Elements (FE) simulations. To reach our objectives we need to develop Numerical strategies for real-time and stable simulation of biomechanical models for needle insertion. Indeed, numerical models of needle-tissue interactions will be used in the control loop of the robot and thus must guarantee at the same time both stability and accuracy, while decreasing the computational expense. Thus, to meet the real-time requirement for robotic control, we need to develop advanced solvers capable of exploiting time coherency and space coherency (domain decomposition methods) and parallelism to solve constrained problems identified as a bottleneck in current simulators.
More information about this subject can be found here.