Team Bunraku

Overall Objectives
Scientific Foundations
Application Domains
New Results
Contracts and Grants with Industry
Other Grants and Activities

Section: New Results

Representations of objects in virtual worlds

New trends in collision detection performance

Participants : Valérie Gouranton [ contact ] , Bruno Arnaldi, Quentin Avril.

Virtual reality environments become more and more complex (in terms of number of geometric entities to handle) and as a result real-time interactions may no longer be satisfying due to the cost of collision computations. Our goal is to propose an efficient environment for virtual reality applications using the efficiency of the hardware. More precisely the focus is made on mapping collision detection algorithms on the hardware layer. Our field of study is on the border between virtual reality and hardware architecture.

During the 1st year of this Phd, we achieved a survey article [42] that presents new trends in collision detection performance. Through this article we presented the new link-up between virtual reality applications (more precisely collision detection algorithms) and computer architecture. More recently, we have presented a first way to parallelise the well-known Sweep and Prune algorithm on a multi-core architecture [65] . We obtained a 3x-4x general speed-up on octo-cores architecture. We proposed a parallelisation of the broad-phase in order to know what kind of result we can expect and also to have a better view of the overall pipeline.

Biomechanics modelling and physical simulation

Participants : Georges Dumont [ contact ] , Maud Marchal [ contact ] , Charles Pontonnier.

Biomechanical model of fracture for deformable objects

In the context of medical simulation, we proposed a novel approach for simulating soft tissue tearing, using a model that takes into account the existence of fibers within the tissue. These fibers influence the deformation by introducing anisotropy, and impact the direction of propagation for the fracture during tearing. Our approach is used for simulating, in real-time, the deformation and fracture of anisotropic membranes, and our method is illustrated with the simulation of capsulorhexis, one of the critical steps of cataract surgery.

This work has been published in the International Conference Medicine Meets Virtual Reality [40] .

Simulation of the interactions between flexible tools and deformable objects

The physical simulation of the interactions between thin and flexible tools with deformable objects is still a challenge, especially in the context of medical applications. Hence, we proposed a new modeling method for the insertion of needles and more generally thin and flexible medical devices into soft tissues. Several medical procedures rely on the insertion of slender medical devices such as biopsy, brachytherapy, deep-brain stimulation. In our model, the interactions between soft tissues and flexible instruments are reproduced using a set of specifically defined complementarity constraints. Each constraint is positionned and applied to the deformable models without requiring any remeshing. Our method allows for the 3D simulation of different physical phenomena such as puncture, cutting, static and dynamic friction at interactive frame rate. To obtain realistic simulation, we parametrized the model using experimental data. We validated our method through a series of typical simulation examples and new more complex scenarios.

This work has been published in the 12th International Conference on Medical Image Computing and Computer Assisted Intervention [48] .

Muscle forces estimation from motion capture data

One of the major preoccupations in industry is the improvement of the working conditions. The goal of this study is to use motion capture data in order to obtain muscles forces involved in the human forearm and hand. The goal is to estimate in real-time the muscle forces involved in several working tasks in order to improve the working conditions. Major interrogations are related to the physical validity of the adapted motions and the correct use of the computed forces and torques for producing physically valid motions.

Our current method to estimate muscle forces follows a four-stepped process:

  1. acquisition of real motion data when following a prescribed scenario. This will lead to obtain a test database for motions that may occur in a working situation;

  2. proposition of a kinematical model of the forearm with a special focus on the elbow. The real data will be mapped on this model to obtain the geometrical data related to the real human who has performed the prescribed task. This model is used to perform an inverse kinematics study to obtain the kinematical parameters related to this motion [58] ;

  3. inverse dynamics study of the forearm that leads to the joint torques related to the motion. This is achieved by building a dynamical model. The mass and inertia parameters used for this point are obtained by methods based on the literature;

  4. use of these preliminary results as an input to a muscular model allowing to access to the forces that are developed inside the muscles. The current method proceeds with an optimization under non-linear constraints algorithm. Antagonism between muscles is taken into account by adding an additionnal constraint that translates the physiology of the joints [26] .


Logo Inria