Multibody & Mechatronic Systems

Mesoscopic cable bundle modelling

Funding: Fraunhofer Institute ITWM

Researcher: Armin Bosten

Advisor: Olivier Brüls and Joachim Linn

Date: Since 2019

Description:

Numerical methods for the mesoscopic simulation of wiring harnesses

Funding: THREAD (European Training Network)

Researcher: Irfan Haider

Advisor: Olivier Brüls

Date: Since 2020

Description:

Nonlinear rod models for system-level braiding process simulation

Funding: THREAD (European Training Network)

Researcher: Indrajeet Patil

Advisor: Olivier Brüls

Date: Since 2020

Description:

Motion analysis of athletes using IMUs

Funding: Teaching assistant

Researcher: Laura Prijot

Advisor: Olivier Brüls and Cédric Schwartz

Date: Since 2019

Description:

Modelling, experimental identification and control of soft robots

Funding: Robotix Academy (Interreg project of the Greater Region)

Researcher: Olivier Devigne, Alejandro Cosimo

Advisor: Olivier Brüls

Date: Since 2019

Description:

Grasp quality metrics for the robotic manipulation of lightweight components

Funding: AACOMA (Interreg project of the Euregio Meuse-Rhine)

Researcher: Louis Dehaybe

Advisor: Olivier Brüls

Date: Since 2020

Description:

Intelligent Robotics for Textile Manipulation

Funding: FRIA grant

Researcher: Norman Marlier

Advisor: Gilles Louppe and Olivier Brüls

Date: Since 2018

Description:

The objective of the research project is to develop efficient and robust planning techniques for the robotic manipulation of textiles (and similarly deformable objects), based on an original combination of finite element modeling methods and machine learning algorithms.

Robot programming based on human motion analysis with IMU measurements

Funding: Robotix Academy project (Interreg, Greater Region) Researcher: Robin Pellois Advisor: Olivier Brüls Date: Since 2017 Description: Human-robot cooperation is an emerging paradigm which aims at combining the complementary skills of a human operator and an industrial robot in order to achieve complex tasks. The human operator has superior ability to analyse a situation, make a decision, plan the action and coordinate the motion in complex and unstructured environments.

On the other hand, an industrial robot can be extremely performant in terms of motion accuracy, reproductibility, reactivity and load carrying capacity. Improvements in this field yield to many new situations. Human and robot can now share the same work space, the robot can see the human, the human can manipulate the robot and other kind of interaction are possible. This new field of robotic also bring new automated programming methods consisting of using information from the operator and/or the environnement in order to make the robot modified his programm by itself. In that way programming robot become easier, faster, more intuitive and no high technical skills are required.

Inside the category of the automated robot programming, one of the most common field is Programming by Demonstration (PbD). The source of inspiration of this field take place in the human-human interaction such as speaking, touching, showing, etc. In the problem of showing something with hands and arms to the robot, the first question to ask is how to acquire information about human motion. Inertial Measure Unit (IMU) sensors appear as a good option. They are wireless, easy to use and doesn't require any special environnement like the opto-electronic systems. They also are lighter, less invasive, smaller than exoskeleton. This work try to demonstrate that IMU can be use to measure the human motion in order to command and control an industrial robot arm.

Modelling methods for human gait analysis

Funding: Teaching assistant at the Aerospace and Mechanical engineering Department of the University of Liège Researcher: Romain Van Hulle Advisor: Olivier Brüls Date: Since 2014 Description: If biomechanical models have been widely used in order to evaluate joint torques and muscle efforts in an inverse dynamics approach, the predictive simulations are far less documented. This means that they cannot be exploited to predict gait pattern modifications when certain parameters are changed (e.g., muscle strength weakened or specifically trained, a joint mobility alteration, hip or knee joint replacement, or a neurological drug treatment). The main objective of this project is to develop predictive modelling tools for gait motion analysis. For that purpose, we model not only the biomechanics of the musculo-skeletal system but also the neurofeedback, which is necessary to simulate a stable gait. Such models could be used to better understand the influence of some rehabilitation programs, medical treatments or disease evolution on the gait.

To achieve this, the study is carried out in two parts: experimental and numerical. The experimental study of gait takes place at the Laboratory of Human Motion Analysis at the University of Liège. The experimental results will allow us to identify the key biomechanical parameters of the numerical model, such as joint torques and muscular efforts. Modelling numerically the neural feedback is of major importance for the development of predictive models. A development of a closed-loop neuronal motor controller, coupled with a biomechanical model is thus our main objective.

Simulation of nonsmooth mechanical systems with frictional contact

Funding: M4 (Walloon Region, Mecatech Cluster) Researcher: Javier Galvez Buezo, Alejandro Cosimo Advisor: Olivier Brüls Date: Since 2015 Description: The frictional contact problem is a phenomenon that appears in a wide range of industrial applications, being of great interest in the automotive or aeronautic industries among others. Various numerical strategies are available to simulate the dynamics of multibody systems with contacts, such as regularization techniques and nonsmooth techniques. The regularization techniques are the common solution used in the industry. However, these have some limitations, for example the constraint violation at the position level. This project focuses on the nonsmooth techniques. Our main goal is to develop and implement, in a finite element software (Oofelie), a non-smooth method capable of dealing with frictional contact, and flexible bodies.

The proposed method splits the motion into smooth and nonsmooth behaviours, combining second order accurate integration of the smooth dynamics and first order for the nonsmooth contributions. This scheme also has the advantage that the unilateral constraints are imposed at position and velocity levels, preventing any non-physical penetration.  The contact law is of the Newton type, relating velocities before and after the impact by a restitution coefficient, both in the normal direction and in the tangential direction for the frictional case.

Stochastic optimization for aeronautical structures

Funding: ORFI (Walloon Region, Skywin Cluster)

Researcher: Juliano Todesco

Advisor: Olivier Brüls and Maarten Arnst

Date: Since 2016

Description:

Over the past decades, Virtual prototyping (CAD & CAE) and Deterministic Optimization have helped to reduce the design cycle and drastically improved the performance of the products. However, performance degrades significantly as consequence of several uncertainties such as: i) manufacturing tolerances; ii) properties of composites materials; and iii) parameters managing related to the processes; among others.

In fact, even today, numerical simulations are conducted by using a simple set of input parameters, that is, fixed boundary conditions and non-variable geometry. These input parameters are defined by a probability distribution which results in output quantities also characterized by a probability distribution.

The main research goal associated to this field is directly related to the development of new approaches for the stochastic modelling of uncertainties in finite element models. The correct representation of uncertainties is crucial to understand how these propagate through the model and affect the behaviour of the system under study.

Since 2000, aerospace and aeronautical applications have shown a radical change towards the use of increasingly flexible and light structures with the aim of optimizing material, shape, aerodynamic performance, etc; fact that has driven the development of increasingly sophisticated non-linear aeroelastic models. However, aeroelastic tools that incorporate optimization processes are rare. The coupling of OOFELIE finite element package with CFD techniques in a suitable framework setting for optimization purposes is the main objective of this research project.