Contrary to rigid robots, the number of degrees of freedom (dof) of soft robots is infinite. On the one hand, a great advantage is to multiply the actuators and actuating shapes in the structure to expand the size of the workspace. In the other hand, these actuators are coupled together by the deformation of the robot which makes the control very tricky. Moreover, if colliding their direct environment, the robots may deform and also deform the environment, which complicates even more the control.
This project would build on our recent results, that use a real-time implementation of the finite element method to compute adequately the control of the structure. The present results allow to compute, in real-time, an inverse model of the robot (i.e. provide the displacements of the actuator that creates a desired motion of the end effector of the robot) for a few number of actuators and with simple interactions with its environment. However, the design of the robots, as well as the type of actuator used are far from optimal.
The goal of this research work is to improve the control methods especially when the robot is in interaction with its environment (by investigating feedback control strategies and by increasing the number of actuators that can be piloted) and to investigate new applications of these devices in medicine (especially for surgical robotics but not only…) and HCI (game, entertainment, art…).
BIOMECHANICAL DEFORMABLE MODELS FOR REAL-TIME COMPUTATION
The accurate modeling of deformations holds a very important place in medical simulation. A large part of the realism of a simulation, in particular for surgical simulation, relies upon the ability to describe soft tissue response during the simulated intervention. In addition, for simulating some procedures, like for interventional radiology, the surgical instruments (needle, catheter, coils, stents…) must also be modeled as deformable. However, obtaining accurate model of the deformation in real-time is a big challenge !
Several approaches have been proposed over the past fifteen years, usually based on the theory of elasticity and the finite element method (FEM).These approaches are often limited to linear models that can not capture deformations with large displacements. Our research aims at obtaining robust models and algorithms for solving the deformations that we observed during a surgery. This research is helped by the development of SOFA and has led to a framework of several FEM models adapted to their geometrical support (curve models, surface models, volume models). Our models are primarily built on corotational approaches, which are popular in computer graphics, in order to deal with the geometrical non-linearities of the deformations. For time integration, we mainly rely on implicit schemes, which remain stable with large time steps but ask for more intensive computation.