Two Grants for Scientific Initiation were awarded in 2012 , for a 6-month period, and they were initiated on the 1st of February, 2012.
Grant holder: Carlos Faria (Master in Biomedical Engineering)
Project title: Trajectory planning and control of a robotic arm for deep brain stimulation
Supervisor: Wolfram Erlhagen
Summary: The objective of this project was to develop an initial approach of a robotic system to play an assistive role in Deep Brain Stimulation (DBS) stereotactic neurosurgery. The robot is expected to position and to manipulate several surgical instrumentations in a passive or semi-active role according to pre-operative directives and instructions of the medical team. We studied and attended standard DBS surgeries leaded by Dr. Manuel Rito at Coimbra University Hospitals, to understand what tasks could be accomplished by the robot. Upon determining the desired robotic system characteristics for DBS surgery, we selected industrial robotic manipulators that fit the basic requirements. The geometric and differential kinematics equations were developed for each robotic manipulator. To test the kinematics equations and the control application in a virtual operating room environment, we used the CoopDynSim simulator. We introduced the serial manipulator concept and implemented the selected robots with several hardware modules. We designed a control application to manoeuvre the robot and devised an initial interface towards positioning/manipulation of instrumentation along surgical trajectories, while emphasizing safety procedures. Although it was impossible to assess the robot’s precision in simulation, we studied how and where to place the manipulator to avoid collisions with surrounding equipment without restricting its flexibility.
Grant holder: Miguel Palhas (MSc in Informatics Engineering)
Project title: GPU implementation of finite volume schemes
Supervisor: Stéphane Louis Clain
Summary: The goal of the project is to achieve an implementation of several finite volume schemes for one or two-dimension geometries and to addapt the GPU framework to the numerical schemes to increase the computational power and reduce the computational cost. We apply such a technology to the concrete problem of poluant transport with 1D or 2D meshes. In a first approach, we have develop a polynomial reconstruction strategy to improve the local approximation of the solution and provide a better accurarate solution. The second step is dedicated to the limiting procedure where we have consider two radically different methods: the a priori method like the MUSCL technique and the posteriori method like the MOOD technique. Both of then have been implemented and tested to check the ability of the method to provide stable and accurate approximation. The last part of the projet is dedicated to the CUDA programation of the first-order scheme for two-dimensional geometries and we manage to get a speed-up of 40 and more.