Note: This content is accessible to all versions of every browser. However, this browser does not seem to support current Web standards, preventing the display of our site's design details.

  

IfA Open House

Back
Abstract:
On Monday, November 30th 2015, the Automatic Control Laboratory (IfA) would like to welcome all students interested in automatic control to the IfA Open House.
The event starts at 4.15pm in ETF E1 with an introduction to the different groups in our lab given by the four IfA professors: John Lygeros, Manfred Morari, Florian Dörfler, and Roy Smith. Following the introduction, there will be different stations where current research topics are presented. We would like to encourage you to take this opportunity to get in touch with members of IfA and find interesting semester and master projects. The event ends with a closing apero in ETL K25.

Time Schedule
4:15 - 5:00pm: presentations by professors (ETF E1)
5:00 - 6:15pm: various stations showcasing the current research topics (logistics to be explained on the day)

Research Station Descriptions
The following gives a short description of the stations that will be showcasing the current research topics from the lab.

Control of Electrical Drives and Compressors

The optimization of the operations of electrical drives is a topic of paramount importance when considering energy efficiency of industrial plants. We present recent work in advanced models for power losses in induction motors, based on magnetic modeling. Optimal control operation can further improve the efficiency of the energy transformation.
More information available here


Autonomous Sailing

The autonomous sailing project focuses on modelling, control and optimisation of autonomous water vehicles. Our team participated at world robotic sailing championship 2015 and won the micro-sailboat class. Next semester and master projects will tackle optimal path planning and sensor fusion of autonomous sailboats. We further plan to extend our fleet and look into hydro-foiled boats.
More information available here


Game Theory for Large Systems

An aggregative game describes a population of systems which influence each other through the population average strategy and aim at selfishly maximizing their individual payoffs. Our research focuses on decentralized algorithms to find population equilibria, by making use of mathematical optimization and fixed point theory. Relevant applications are the charging coordination of a fleet of plug-in electric vehicles and the shortest-path problem in a traffic congested network.
More information available here


RoboCup, the Z-Knipsers

RoboCup promotes research in the fields of robotics and artificial intelligence through the game of soccer. The objective is to have a match between robots and the FIFA world cup champion by 2050. Our team, the Z-Knipsers, participated this year at the Iran and German Open as well as at the World Cup in China. Spring and summer 2016 will be again filled with competitions and we are looking for motivated students to tackle problems in the domains of self-localization, ball detection and team behavior.
More information available here


Embedded Control Systems

Industrial control algorithms have to run on embedded platforms in real-time. To control complex systems as well as to improve efficiency, state-of-the-art control schemes are based on optimization techniques. Technological advancements in embedded processing and the availability of cheap but powerful micro-controllers enable new, powerful optimization methods. These intelligent algorithms dwarf the abilities of conventional controllers, while still being able to run in a real-time setting. Applications range from power electronics to trajectory planning for quad-rotors.
More information available here


Airborne Wind Energy

The aim of the airborne wind energy project is to achieve fully-autonomous flight of kites for power generation. Such kite systems exhibit highly nonlinear dynamics and are limited to ground-based actuation and sensing. To tackle these challenges we offer Semester and Master projects in system identification, modelling, estimation and control of kites.
Most projects provide a balance between hardware and theoretical development and consist of a strong experimental component during field tests at different locations in Switzerland.
More information available here


Active control of thermoacoustic machines

Thermoacoustic machines convert thermal energy into sound waves which in turn can drive a thermoacoustic cooler or can be directly converted into electrical energy using an electrodynamic transduction mechanism. These thermoacoustic machines have a high potential for the development of clean, sustainable and alternative energy systems by utilizing any of: low temperature waste heat recovery, biomass and gas combustion heat, or solar energy, respectively. A demo will be given to show the control of a thermoacoustic cooler machine and real-time digital control of thermoacoustic instabilities inside a Rijke tube.
More information available here


Distributed Smart Energy Systems

The electricity grid is one of the backbones of modern societies. Every year multiple trillion Dollars worth of electricity are consumed globally. However, the ongoing transformation of the energy system towards renewable sources puts the reliability and security of supply at risk. We investigate these new challenges from a theoretical viewpoint. Our goal is to identify conditions that guarantee robust and efficient operation of power systems at different conceptual levels and to develop control algorithms that reach these favorable operating states.
In our semester and Master projects we offer students the opportunity to learn about large-scale infrastructure systems and the tools to analyze them, such as optimization and control over networks, graph theory and non-linear system dynamics.
More information available here


Energy Efficient Buildings and Districts

Building energy management is an active field of research as the potential in energy savings can be substantial. The energy supply options have increased with the introduction of renewables and the possibility of cooperative energy management within energy-hubs. The energy hub houses the expensive but energy efficient equipment that is shared by the building community, and provides the opportunity for load shifting operation among interconnected buildings. Significant efficiency gains can be envisaged but require sophisticated control systems that can handle the variability in energy prices, supplies and reliability in the presence of operational uncertainties, such as weather and occupancy.
More information available here


Data Driven Control

Data Driven Control uses ideas from Data Driven Optimization and Machine Learning to learn to control a plant solely based on experiments on the real plant or simulations of it. Its main advantage is that it is a black-box approach which does not require an (a-priori) model of the system. Hence, it promises to save effort on system identification and controller tuning. It might also prove advantageous for systems for which no analytic model is known, model parameters are highly uncertain or for which models exhibit features that prevent the usage of traditional tools for controller design. At IfA, we are currently applying these methods to Big Data applications, such as traffic control and credit card fraud detection. We are also interested in investigating theoretical properties of these methods.


Chemical Process Control

Crystallization is an often-employed separation step in the chemical industries. The size and shape of the resulting particles determine important properties such as tabletability or bioavailability. Unfortunately, the understanding and consequently also the control of particle size and shape has been limited, mostly due to the unavailability of fast and reliable measurement techniques. The separation processes laboratory at MAVT has developed a stereoscopic imaging setup (cf. Figure 1) that allows for the measurement of these features and – together with IfA – has demonstrated the applicability of the setup for process characterization, laying the foundation for model-based process control.
More information available here


Stochastic Control

When designing controllers for real-world applications, one is often confronted with two difficulties: first, systems are uncertain and influenced by disturbance; second, systems are high-dimensional and distributed. In the control design process, both issues need to be accounted for. At this station, we will give a brief overview of methods addressing these issues, and present existing projects.
More information available here


Distributed Craziness

The “Distributed Craziness” project aims at the real-world implementation of distributed estimation and control algorithms for a fleet of nano-quadcopters: the “Crazyflies”. The focus is to build a system that truly embodies the complexities of distributed estimation and control. Therefore, each agent is able to take measurements with on-board sensors, make decisions via on-board computation and communicate via an ad-hoc network. The next steps in the project are to further develop the distributed architecture of the flying agents testbed and to replace the camera-based Vicon system for tracking by an ultra-wide-band based position measurement system in order to maximize autonomy of the agents. As for the distributed control and estimation algorithms, the goal is to further fuse together cutting edge theoretical work and challenging robotics problems.
More information available here


Autonomous Racing

The ORCA (Optimal RC Racing) Project developed (and improves) a test bed consisting of a race track, a infrared camera based tracking system and modified 1:43 dnano RC cars, in order to study control algorithms allowing high-speed, real-time control. On the test bed different fast MPC algorithms are implemented, allowing the cars to online plan their trajectory only based on the track layout and avoid other cars. To allow competitive racing between automatically controlled cars, game theoretical methods are used to derive new control strategies, suited for competitive racing.
More information available here


Type of Seminar:
Public Seminar
Speaker:
organized by the IfA laboratory
Date/Time:
Nov 30, 2015   16:15
Location:

ETF E1
Contact Person:

No downloadable files available.
Biographical Sketch: