Robust control of state/input-constrained robotic systems: An LMI approach
This Keynote Lecture presents my recent works on the robust control of some robotic systems
subject to state and/or input constraints. The LMI approach is introduced for the design of the stability
conditions of the closed-loop system subject also to parametric uncertainties and external disturbances. Some
robotic systems, on which I am actively working, are first presented. After that, I present three different robotic
systems, namely the Inertia Wheel Inverted Pendulum, the simplified helicopter model and the two-sided 1-
DoF impact mechanical oscillator,on which the design approach was applied.Some technical lemmas used for
the design of LMI conditions that are obtained from BMIs andvwhich are derived from a set of constrained
stability conditions,are also presented. Some simulation results are given to show the robust stabilization of
these considered robotic systems and also the interest in considering the state and input constraints in the
design of the LMI stability conditions and then of the control law.
Sliding mode control of robotic systems for rehabilitation
The basic idea consists to control two degrees of freedom exoskeletons, at the hip and knee
joint levels, as well as the variation of kids parameters. These controllers are developed in order to make the
exoskeletons track the desired trajectory so to allow children to increase their brain plasticity by an intense and
precise work on the gait scheme. The proposed controller is based on the sliding mode approach. Adding an
integral action to the control, performances can be improved. Then, adaptive approaches estimating parameters
online presents the best performances.
Performance of Photovoltaic (PV) Systems
The performance of the system, which mainly concerns the energy output, both in quantity and in timing, depends on the operating conditions and the detailed configuration of the system. In turn, the operating conditions depend on the location of the system, which governs the solar irradiation received, the ambient temperature and other climate-related aspects that influence the system performance. How the system performs determines its technical and economic feasibility. This lecture focusses on how to measure and assess the performance of the components of the PV systems and on how to define and address the performance of the PV system in the field. PV systems performance deals with the performance of the PV module, which performs the actual conversion process including measurement of the module rating, prediction of energy yield and changes to module performance in operation. Performance of the PV system includes the balance of system components, prediction and monitoring of PV system output and the environmental assessment of the systems. Finally, PV performance considers individual system types, including grid-connected and stand-alone system configurations, concentrator systems and specific performance aspects relating to building integration.
Electrical Energy Storage and its Advanced Role in Power Electrical System and Micro-Grids
The integration of Renewable Energy Resources, in particular solar PV, into Power Electrical System and Microgrids can cause several impacts, due to their intermittent characteristics. These impacts on the grid operation may include voltage variations, frequency variation, voltage unbalance, stability, protection and challenges for managing… These effects are even more complicated for islanded or weak grid areas. Solutions by using Electrical Energy Storage (EES) can significantly reduce these impacts and can contribute ancillary services for power system. This keynote presents experiences of CEA-INES on modelling of ESS and also solutions by applying intelligent control and energy management of Electrical Energy Storage systems for:
- Frequency control
- Voltage control
- Congestion management
- Guarantee power
- Stability improvement …
Application of deep learning in Arabic
speech processing and recognition
The presentation will give an overview of the recent advances in the applications of deep learning in language processing in general and speech recognition in particular. The talk will then provide a short review on the challenges facing Arabic Language and the recent research in Arabic language and Arabic speech. Finally we present our research activities along several Arabic language applications including, transcription of Arabic news, speech recognition of
Arabic dialects, Arabic language modeling using RNN, Arabic phoneme recognition, speech segmentation, automatic restoration of diacritical marks of Arabic text, Quran recitation, on-line Quran annotation, recognition of poem meters, and provide an overview of Arabic conversational agent.
The Planck Balance – calibration of E2 mass standards using the new definition
The Planck Balance (PB) is an electronic mass comparator, which allows the calibration of weights in a continuous range from 1 mg to 100 g using a fixed value of the Planck constant, h. It uses the physical approach of Kibble balances that allow the Planck constant to be derived from the mass.
Using the Planck-Balance no calibrated mass standards are required during the calibration of mass standards any longer, because all measurements are traceable via the electrical quantities to the Planck constant h, and to the meter and the second. This allows a new type of balance after the redefinition of
the SI-units on 20th of May 2019. In contrast to many scientific oriented developments of Kibble balances, the Planck Balance is focused on robust and daily use. The Planck Balance will allow relative measurement uncertainties comparable to the accuracies of class E2 mass standards, as specified in OIML Recommendation R 111-1. The Planck Balance is developed in a cooperation of the Physikalisch-Technische Bundesanstalt (PTB) and the Technische Universität Ilmenau in a project funded by the German Federal Ministry of Education and Research.
Standardization in Wireless Communication
Potential of carbonaceous materials for high performance sensors
Nanocomposites films based on carbonaceous materials and polymers provide an innovative approach for the realization of scalable flexible sensors with a high performance and low manufacturing costs. The realized sensors can overcome several limitations of conventional sensors and provide decisive advantages, e.g. adjustable measurement range, high sensitivity, high robustness and scalability. Some of these aspects will be exemplarily discussed for sensors based on carbon nanotubes (CNTs) and Graphene oxide (GO) for measurement quantities including temperature, humidity, strain and pressure. The CNT/GO based sensors are highly attractive for applications in the fields of structural health monitoring (SHM), robotics, medical diagnostics, ioT and environmental monitoring.
Quantitative assessment of visual information: how far can we go?
Despite the major advances made in digital image and video processing and their integration in various applications and systems, visual media still suffers from many artifacts that can affect or impair quality as perceived by humans. Such factors can be related to acquisition, processing, compression, transmission, display, printing, and reproduction. This has promoted a growing interest in developing robust objective approaches for evaluating image/video quality. Standards bodies, i.e., ITU-T VCEG and MPEG, now require subjective tests to be performed on all proposals. In this talk, we provide an overview of the field of visual quality assessment including past efforts, current trends, challenges, and future directions. For subjective visual quality assessment, a short review will be given on standard subjective evaluation procedures. For objective visual quality assessment, its applications, the different techniques used in image and video quality metrics will be discussed. The general framework used by the classical HVS-model based image quality metrics (IQMs) will be briefly introduced first, followed by some new approaches which do not rely on the HVS models, e.g., the SSIM and HOSVD. For video quality assessment, we discuss how temporal/motion information is used to measure the subjective video quality is considered in developing quality metrics. While we will briefly outline full reference techniques, we will focus more on the challenge cases of no reference based approaches.
Robot Operating System (ROS): How did it revolutionize robotics software development?
Robot Operating System (ROS) becomes nowadays the de-facto standard for developing robotics applications. The pre-birth initiatives were emerging from STanford AI Robot (STAIR) project and Personal Robots (PR) program, which aimed at creating dynamic software systems for robotics applications, until 2007 when Willow Garage, a major robotics investor, boosted the development of this initiative and contributed to the release of the first ROS software packages in 2009. The first version of ROS was released in 2010 and nowadays ROS becomes the largest ecosystem and platform for robotics software development. In just a few years of its release, ROS has witnessed a huge community with increasing number of users and developers from academia and industry, as well as hobbyists. How did ROS revolutionize robotics software development in just a few years?
In this presentation, I will give an overview of ROS and its evolution in the past years after its release. I will unveil the secrets of ROS that makes it a revolutionary solution for developing robotics applications. I will share my experience, as a computer scientists working on robotics, with developing robotics applications in the pre-ROS and post-ROS times, and how ROS made a complete shift in the software engineering and development approaches for mobile robots. The presentation will also give a small overview of the main concepts of ROS and the most important libraries and packages that comes with it. Video demonstrations and real illustrations will be presented.