Student Projects

Projects are extracted from ISA database, some delays may occur. For more information about the projects, send an email to the project “contact” person.

For registration please contact the secretary of lab Mme. Bouendin.

For Internship + PDM in indusry, please contact Dr. Alireza Karimi

Directives (2014) for projects at LA

Please note that project status (available/taken/etc) may not be accurate, please contact the designed “contact” person for the actual status.

Information to add/manage projects on ISA can be found here  (not official).

LA projects on ISA (Jones, Ferrari Trecate, Karimi, Salzmann)

ALT

Houses are commonly equipped with PV panels, electric batteries, variable power heat pump and electric vehicles. This means single homes are becoming complex energy hubs which require an efficient energy management system (EMS) that can be configured in a simple way while satisfying complex needs such as improving self-consumption and reducing cost.

The energy systems group at CSEM has developed the software NRGMaestro(TM) a next-generation energy management system that is based on Model Predictive Control (MPC) and which can be easily reconfigured through an automated controller synthesis process. This energy manager has been embedded in the commercial product crystalball.solar [https://crystalball.solar/] and is currently in operation in several buildings in Switzerland.

One important lesson learnt from these prototype deployments is the need to deal with changes of behaviors of the system or house occupants. MPC utilizes a model of the building that is based on an initial evaluation of the main physical characteristics of the building. When significant modifications in occupant behavior occur ( e.g. opening blinds, change of season for heat pump, hardware upgrade ), the model fails to accurately predict what will happen and is currently unable to adapt to these prediction errors. The goal of the internship is design and implement a closer integration of the data collected on the building and the model used in the optimization.

Different methods will be reviewed and compared on a theoretical and practical standpoint, including

o State and disturbance estimation

o Fault detection with change point algorithms

o Machine learning applied to dynamical systems, including system identification, and learning MPC

Tests will be conducted using data collected on the real test sites

While the objective of the work is clearly industrial, there will be scope to present the results in scientific journals or at conferences. The student will be part of a team of twelve experienced engineers and researchers and will face actual practical challenges of software development for energy applications.

Requirements:

An interest in energy systems and energy management problematics.

Familiarity with some of the following topics is a plus:

o System identification

o Machine learning algorithms and good practices

o Kalman filtering

o Model predictive control and numerical optimization

Interest and experience in programming ( Python )

Comment
Contact: Tomasz Gorecki, [email protected]

Type of project: Master in industry or internship

minimum 5 months

Location: Neuchatel

Professor(s)
Colin Neil Jones (Laboratoire d’automatique 3), Christophe Salzmann
Administration
Nicole Anne Bouendin
External
[email protected]

TESA Technology is a Swiss division of Hexagon AB, a worldwide group leader in sensors, software, digital factory and digital world, as well as in autonomous solutions. In particular, TESA manufactures precision instruments for Hexagon’s Coordinate Measuring Machines (CMMs), notably probes and articulated motorised probe head for orientating these probes near the workpiece to be measured. Despite slotted brushless motors are cost-effective and efficient solutions for motorising devices, these motors produce, at slow rotational speeds, cogging torque that compromises its use in various measuring solutions.

This project targets the design of an anti-cogging driving unit of slotted brushless motors for articulated probe heads and probe’s rotatable joints. The project’s milestones comprise:

1) study of the technical background of slotted brushless motors, and definition of motor requirements for use in metrology

2) design and simulation of most-promising driving control unit

3) realisation of a functional prototype

4) validation and discussion of the realised functional prototype.

Professor(s)
Alireza Karimi
Administration
Isabelle Stoudmann Schmutz
Site
ddmac.epfl.ch
ALT

The EPFL babyfoot is under continuous improvement.

While the babyfoot can easily intercept the ball, it has difficulty to capture it and make a pass. Juggling with the ball is another challenge. This project aims at improving the ball handling to permit capture and juggling, and then shoot toward opponents goal.

Students suggested improvements are also welcome.

Professor(s)
Christophe Salzmann (Laboratoire d’automatique 3)
Site
https://www.epfl.ch/labs/la/studentprojects/babyfoot/
ALT

The optimal management of distributed renewable energy ressources requires an increasing number of advanced hierarchical controllers in distribution grids. However, these controllers require knowledge of the network’s parameters, which can be identified using voltage and current measurements. In order to boost the accuracy of the identification, any available knowledge about distribution grids needs to be used. This can be done through Bayesian a-posteriori estimation, and the definition of suitable prior probability distributions. So far, our research focused on single phase networks. However, real-world power systems are mostly three-phased, and existing priors need to be adapted and generalized. This project consists in understanding Bayesian identification as well as the prior structure used for single phase distribution grids, to write the equations of new, three-phases priors, and implement them into the existing simulation tool.

Skills involved:

– Modelling and simulating power systems

– Python programming on a remote server

– Excellent skills in linear algebra, in particular for large sparse matrices

Comment
Assistant: Jean-Sébastien Brouillon ([email protected])
Professor(s)
Giancarlo Ferrari Trecate, Jean-Sébastien Hubert Brouillon
Administration
Nicole Anne Bouendin
Site
https://la.epfl.ch/pi
ALT

A large class of systems can be described by a linear but infinite dimensional systems. Example of such systems are LTI models with time delay, or systems governed by (linear) partial differential equations. The goal of this project is to implement a novel method to design continuous time controller, and validate it in simulation on two different systems.

Required courses: Advanced control systems

Required: MATLAB, some exposure to numerial optimization.

Comment
contact: Philippe Schuchert
Professor(s)
Alireza Karimi
Administration
Isabelle Stoudmann Schmutz
Site
https://www.epfl.ch/labs/ddmac/
ALT

Increasing complexity in modern cyber-physical systems challenges classical model-based control techniques that rely on low-dimensionality parametric representations of the system dynamics. Data-driven control, on the other hand, aims to directly design a control law leveraging the availability of process data, without explicitly identifying a system model.

In particular, the successful application of a data-enabled predictive control to optimal trajectory tracking for unknown systems has recently fostered a renewed interest in the implicit system description offered by behavioural theory. Nevertheless, it is still unclear which are the advantages and the pitfalls of this representation over the ones obtained through state-of-the-art system identification approaches.

The goal of this project is to compare the prediction ability of behavioural and auto-regressive models and to assess whether non-linear numerical optimization can be used to optimally choose the predictor in the presence of noisy input-output data. The performance of the developed algorithms will be evaluated through simulations of possibly complex, non-linear or time-varying, systems.

Requirements: Basic knowledge of control theory and linear algebra, familiarity with MATLAB.

Comment
Assistant: Martin Andrea
Professor(s)
Giancarlo Ferrari Trecate, Andrea Martin
Administration
Nicole Anne Bouendin
Site
https://la.epfl.ch/pi

Model Predictive Control (MPC), which anticipates future events and adopts control actions accordingly, is popular in industry because of its compatibility with system constraints and ability to deal with undesirable randomness. However, it usually relies on empirical models obtained by System Identification (sysID), which requires considerable expertise. Currently, DeePC, based on behavioral models and able to avoid sysID, is considered as a variant, easier to implement and also showing a better performance.

Although there exist results on its stability and robustness, little attention is paid to noise attenuation within the DeePC scheme. Since no traditional models, like state-space model, are available, one cannot directly adopt classical filtering methods. Besides, data are explicitly shown in behavioral models, which may help us bring up better filtering or adaptive control strategies than existing ones.

In this project, we plan to look into a gradient method that recursively modifies behavioral models, tune the scheme and test its performance.

Contact: Baiwei Guo, [email protected]

Comment
Assistant: Baiwei Guo, [email protected]
Professor(s)
Giancarlo Ferrari Trecate, Baiwei Guo
Administration
Nicole Anne Bouendin
Site
www.epfl.ch/labs/la/pi/

Model Predictive Control (MPC), which anticipates future events and adopts control actions accordingly, is popular in industry because of its compatibility with system constraints and ability to deal with undesirable randomness. However, it usually relies on empirical models obtained by System Identification (sysID), which requires considerable expertise. Currently, DeePC, based on behavioral models and able to avoid sysID, is considered as a variant, easier to implement and also showing a better performance. Although there exist results on its stability and robustness, little attention is paid to noise attenuation within the DeePC scheme. Since no traditional models, like state-space model, are available, one cannot directly adopt classical filtering methods. Besides, data are explicitly shown in behavioral models, which may help us bring up better filtering or adaptive control strategies than existing ones. In this project, we plan to look into a gradient method that recursively modifies behavioral models, tune the scheme and test its performance.

Comment
Assistant. Baiwei Guo, [email protected]

(copy of project 10253)

Professor(s)
Giancarlo Ferrari Trecate, Baiwei Guo
Administration
Nicole Anne Bouendin
Site
www.epfl.ch/labs/la/pi/

This project aims to design, identify and control a miniature hovercraft for an air hockey table.

Comment
Assistant: Roland Schwan
Professor(s)
Colin Neil Jones (Laboratoire d’automatique 3), Christophe Salzmann
Administration
Nicole Anne Bouendin
ALT

Deep networks have become invaluable tools for supervised machine learning, e.g., they have been successfully applied in many computer vision and classification tasks. While often offering superior results over traditional techniques, deep architectures are known to be challenging to design and train such that they generalize well to new data. Important issues with deep architectures are numerical instabilities in learning algorithms commonly called exploding or vanishing gradients.

This project aims to develop a novel energy based classifier for classification purposes by employing Hamiltonian based deep neural networks (H-DNNs). These neural networks stem from the discretization of ordinary differential equations and guarantee non-vanishing gradients by design. Hamiltonian systems are attractive due to their energy-preservation properties which can result in more stable deep networks. Recently, HDNNs are successfully implemented for several data-sets.

Comment
Assistants: Clara Galimberti and Muhammad Zakwan ([email protected])
Professor(s)
Giancarlo Ferrari Trecate, Muhammad Zakwan
Administration
Nicole Anne Bouendin
Site
https://www.epfl.ch/labs/la/pi/

The Automatic control lab has recently acquired a new educational platform from Quanser: the Qube Servo-2.

The aim of this project is to first prototype a new set of “caps” that can be attached on top of the Qube, such that the dynamics are more challenging that a simple DC motor. This addition should be designed such that a large variety of dynamics can be obtained, by simply changing some weights.

In a second stage, identification and control of this system will be of interest. A gain-scheduled controller will be designed to stabilise the system, along with an estimator to estimate correct values for the scheduling vector.

Finally, the result will be compared to a robust controller, stabilising all different models simultaneously.

Required courses: System Identification, Advanced Control Systems.

Required knowledge: MATLAB, LabVIEW, some C programming

Comment
contact: Philippe Schuchert
Professor(s)
Alireza Karimi
Administration
Isabelle Stoudmann Schmutz
Site
la.epfl.ch
ALT

Deep neural networks have proved successful in many application domains, such as image recognition, language comprehension, control of dynamical systems and sequential decision making. However, recently, incorporating appropriate inductive bias a-priori in the training phase has shown improvements in capturing the ground truth of the data. Such neural networks are called physics-inspired neural networks.

In this project, our goal is to develop architectures of neural networks that can capture the underlying energy functions of port-Hamiltonian systems (PHs) by using the data. The framework of port-Hamiltonian systems can model a number of real-world applications ranging from simple mass-spring system to the complex DC microgrids. Moreover, learning the underlying physically-consisted model of a dynamical system can subsequently lead to model-based controller design of complex non-linear systems.

Image from paper: “Hamiltonian Neural Networks”, Sam Greydanus, Misko Dzamba, Jason Yosinski . 2019.

Comment
Assistants: Clara Galimberti ([email protected]) and Muhammad Zakwan ([email protected])
Professor(s)
Giancarlo Ferrari Trecate, Christophe Salzmann
Administration
Nicole Anne Bouendin
Site
https://www.epfl.ch/labs/la/pi/
ALT

Background

Model Predictive Control is used in buildings for a number of goals, such as improving energy efficiency and thermal comfort, or providing ancillary grid services. To do so requires measurement and prediction of the variables affecting the temperature in a building. For the weather, professional forecasts are available. Another significant variable is the heat emitted by the human occupants in the building. This variable is both volatile and difficult to measure.

Thesis

We currently use the Polydome for MPC experiments. The Polydome is a one-room building used for classes and seminars in the north-east corner of the EPFL campus. However, no measurement of human occupancy is available at the moment. For the thesis, you will conduct the following steps:

* A literature search on the methods for measuring human occupancy in a building.

* An assessment of those methods for their applicability in the Polydome. Key questions may be the size of the room or privacy concerns.

* The practical implementation of one or two chosen methods in the Polydome.

* Tuning and validating the system by a series of hand-counts of the occupancy and comparison with the sensor data.

* If time allows, an investigation of learning schemes, updating the sensor calibration to account for changes in the system, such as seasonal variations.

Requirements

Basic knowledge of the following topics:

* Sensors and signal processing

* Fluid and thermo dynamics

* Programming in Matlab or Python

Contact

Manuel Koch

[email protected]

Professor(s)
Colin Neil Jones (Laboratoire d’automatique 3), Christophe Salzmann
Administration
Nicole Anne Bouendin
ALT

Smart grids bring about automation to electricity networks by establishing a two-way interactive system between consumers and energy providers through fetching users’ power consumption in discrete time intervals. These fine-grained measurements can be used for inferring models of households’ energy consumption, and for analyzing the effect of various geographical or societal factors on consumption patterns. While there has been a vast amount of work on household electricity prediction, the proposed approaches suffer from requiring long data collection periods before the models can be used and are sensitive to the change of households’ habits or technological renovations.

To alleviate the requisition of big training datasets and improve the flexibility of the models to adapt to the changes in consumption patterns, one can decompose the learning algorithm into two layers: a) learning a generic function describing shared consumption patterns among households and b) training personalized models on top to account for personal differences. This scenario is well-described in the context of meta-learning.

In this project, the student is asked to apply state-of-the-art meta-learning algorithms to the problem of short-horizon prediction of household consumption. This proposal is an applied semester project and requires strong programming skills, plus a good knowledge of machine learning.

Main tasks:

Familiarizing with energy datasets, visualization and exploratory data analysis, identifying relevant meta-learning algorithms, simulation, and experimental validations.

Comment
Assistant: Mahrokh Ghoddousiboroujeni ([email protected])
Professor(s)
Giancarlo Ferrari Trecate, Mahrokh Ghoddousiboroujeni
Administration
Nicole Anne Bouendin
Site
https://www.epfl.ch/labs/la/pi/

Introduction

Concentrated sunlight-powered CO2 reduction to chemicals and fuels via reverse water-gas shift (RWGS)[1] and Sabatier[2] processes offers great promise in closing the carbon loop and accelerating the renewable energy transition. In addition, energy-efficient LED is selected as the light-compensation strategy during solar-poor and dark conditions to ensure continuous operation. Such a hybrid solar/LED powered system concept is illustrated in Fig. 1a.

Due to the transient and intermittent nature of sunlight, dynamic irradiation control strategies are required in order to determine when the LED should be turned on/off under different solar conditions. With a rule-based irradiation control strategy[3], the dynamics of the hybrid powered system have been investigated in Matlab (see Fig. 1b) to help identify the optimal operating conditions under typical solar direct normal irradiation (DNI) profiles. However, this control scheme can lead to temperature overshoot for strongly fluctuating DNI cases when the reactor thermal inertial is low, thus highlighting the need for more sophisticated control strategies[3, 4].

Objective

This project is aimed at designing an optimal solar/LED irradiation control strategy for the hybrid powered chemical system to achieve efficient fuel production while also bypassing the temperature overshoot issue as encountered in the rule-based control scheme.

Such a control strategy should take into account realistic sunlight conditions and LED illuminating characteristics along with other practical constraints set by the reactor, catalyst and chemical processes. Special attention should be paid to understanding the fundamental differences between the exothermic Sabatier process and the endothermic RWGS process that may require distinct control strategies.

The control strategy will be incorporated into the previous Matlab model to identify the optimal operating conditions, thus guiding reactor design and practical operation.

Work plan

* Review of the state-of-the-art control strategies to identify potential options for this project

* Development of a refined dynamic model by incorporating the selected control strategies into the previous Matlab code

* Assessment of the selected control strategies to screen the most optimal choice

* Completion of the project report

References

[1]. Sastre, Francesc, et al. Sunlight-fueled, low-temperature Ru-catalyzed conversion of CO2 and H2 to CH4 with a high photon-to-methane efficiency. ACS Omega, 2019.

[2]. Upadhye, Aniruddha A., et al. Plasmon-enhanced reverse water gas shift reaction over oxide supported Au catalysts. Catalysis Science & Technology, 2015.

[3]. Oldewurtel, Frauke, et al. Use of model predictive control and weather forecasts for energy efficient building climate control. Energy and Buildings, 2012.

[4]. Di, Wu, et al. Modeling and simulation of novel dynamic control strategy for PV-wind hybrid power system using FGS2PID and RBFNSM methods. Soft Computing, 2020.

Professor(s)
Sophia Haussener (Laboratoire de la science et de l’ingénierie de l’énergie renouvelable), Giancarlo Ferrari Trecate
ALT

Real-time pricing (RTP) is a rising concept in smart grids which assigns electricity prices in discrete time intervals. This scheme aims at encouraging consumers to actively manage their electrical equipment, and consequently, balance electricity demand and generation. While several studies have illustrated that the RTP strategy can effectively reduce peak load and flatten the demand curve, an optimal pricing scenario that maximizes social welfare and customer satisfaction while increasing the providers’ revenue is still an open question.

The standing assumption in demand management through RTP is the elasticity of consumption, meaning that the demand responds to the price signal. The consumer’s response is a complex black-box model with an unknown structure. Additionally, it is hard to learn this response through arbitrarily pricing, as it affects both customers’ experience and providers’ income. This project proposes to use Bayesian optimization, a well-suited tool for maximizing hard-to-evaluate objective functions, for electricity pricing.

Power TAC is an open-source platform for simulating electricity retail markets. It is suggested that the student runs experiments on this software to evaluate the designated algorithms and methodology. Other simulation tools or models are welcomed.

Further reading:

Introduction to RTP [https://www.povertyactionlab.org/evaluation/real-time-pricing-reduce-electricity-use-united-states]

Bayesian optimization in Python [https://machinelearningmastery.com/what-is-bayesian-optimization/]

Power TAC [https://www.powertac.org], [https://www.powertac.org/wiki/index.php/Main_Page]

Main tasks:

a) familiarizing with Power TAC platform, b) problem formulation using Bayesian optimization, c) simulating and experimental validations.

Requirements:

Machine Learning, Python programming, experience with Git is a plus.

Comment
Assistant: Mahrokh Ghoddousiboroujeni ([email protected])
Professor(s)
Giancarlo Ferrari Trecate, Mahrokh Ghoddousiboroujeni
Administration
Nicole Anne Bouendin
Site
https://www.epfl.ch/labs/la/pi/
ALT

Optimal control using data is increasing interest of both academic and industrial communities. In comparison with conventional model-based control, data- driven control has the advantage that it can be applied in scenarios where data is readily available, but the system and uncertainty models are too complex to obtain or maintain, e.g., distributed systems (large-scale power systems) or energy efficient buildings, etc. Data-driven techniques are expected to be more adaptive and reactive to changing environment and uncertainties. Usually, a mathematical model of the plant is not available and thus a model needs to be identified from a set of experimental data and then the controller is designed using this identified model. Such a 2-step or indirect data-driven procedure has been proven to lead to potentially sub-optimal controllers. Moreover, the model that best fits the data is not necessarily also the most suitable for controller tuning.

Recently, behavioral modelling (BM) is proposed for the control design of the linear time-invariant systems. This result, known as the Willem’s fundamental lemma, shows that the subspace of input/output trajectories of a linear time- invariant (LTI) system can be obtained from the column span of a data Hankel matrix, thereby avoiding a parametric system representation. Multiple direct data-driven control methods have been proposed based on BM.

The goal of this project is to develop data-driven distributed control of linear time-invariant (LTI) identical dynamically coupled systems known as decomposable systems. These systems result from the interconnection of a large number of identical subsystems, for example, multi-agents, vehicle platoons, etc. If the state-space matrices of these systems satisfy a certain structural property, then it is possible to derive a data-driven procedure by leveraging tools from behavioral modelling for designing a distributed controller which has the same interconnection pattern as the plant.

Comment
Assistant: Muhammad Zakwan ([email protected])
Professor(s)
Giancarlo Ferrari Trecate, Muhammad Zakwan
Administration
Nicole Anne Bouendin
Site
https://www.epfl.ch/labs/la/pi/

High performance parallel implementation of numerical optimal control algorithms.

Comment
Assistants: Roland Schwan, Yuning Jiang, Petr Listov
Professor(s)
Colin Neil Jones (Laboratoire d’automatique 3), Christophe Salzmann
Administration
Nicole Anne Bouendin
ALT

Mikron Automation is the leading partner for scalable and customized assembly systems.

The Polyfeed (https://www.mikron.com/automation-solutions/platforms-systems/feeding-systems/mikron-polyfeed/) is a feeding system able to feed components to be picked by a cartesian robot (X-Y-Z + rotation). The feed and pick processes are based on visual recognition and vertical vibration systems. The feeding speed can reach 80-110 parts/min, depending on the component design and size (varying from 1 to 50 mm3 size). The current trajectory implementation is based on a simple but optimized pick and place movement. A basic obstacle avoidance is implemented, based on delayed displacements across one axis.

Mikron is currently developing a Robotic Cell, which will have a layout similar to the Polyfeed, but will be used to realize more complex operations, as assembly or insertion of components. It therefore becomes strategical to introduce the notion of optimized path planning with obstacle avoidance, imposed waypoints and a target position. Given this information, entered by an operator on the user interface, the system must be able to propose the best trajectory, optimized with regards to speed and vibration aspects. It must also be able to recompute dynamically the trajectory in the case of an unexpected event.

The main activities for this project are:

(1) Familiarization with the existing trajectory algorithms and their implementation on the Polyfeed. (2) Study of existing path planning algorithms and selection of the most adapted ones. (3) Design and simulation. (4) Testing on a functional prototype. (5) Qualification of the trajectories based on measurements.

Comment
This is a PDM+Internship project
Professor(s)
Alireza Karimi
Administration
Isabelle Stoudmann Schmutz

Industrial projects (not yet on ISA)

To address climate change properly and accelerate energy transition from fossil fuels to renewable energy, new energy generation technologies are required. They have to be reliable, efficient, scalable and provide sufficient low-cost energy to satisfy current and future demands. Airborne Wind Energy (AWE) is an emerging technology that involves autonomous tethered aircraft to produce electricity from high altitude winds. 
Skypull can offer a solution and proposes an innovative approach, based on a new revolutionary type of flight device, aimed at maximizing the lift, with minimum resistance increase and allowing autonomous operation: a “box” drone with proles multi-element aerodynamics, characterized by high aerodynamic performance, intrinsic structural strength, low weight and low production costs. 
In order to keep the drone operational over prolonged periods of time it is critical to recharge the onboard electronics during operation. Skypull developed the technology of regenerative differential braking, which allows the controlled braking of the aircraft while simultaneously regenerating energy. 

Goal 

In order to better understand the process of differential braking our drone model needs to be extended to accommodate for the braking process. The models focus will be on the drivetrain (motor controller, motor and propeller), but should be integrated in the drone model. 

Based on this model a set of control laws will be developed to enable the drone to brake, turn and regenerate energy in a stable and robust way.

Responsibility 

* Literatue/Skypull technology review 
* Installation of a test bench 
* Implementation of the drive train model and integration into drone model 
* Derivation of control strategy for differential braking 
* Testing on Skypull hardware

Requirements 

* Good understanding of Simulink / MATLAB 
* Good fundamental knowledge in control system design and modelling 
* Basic knowledge of aerodynamics and aircraft actuators 
* Basic knowledge of motor controllers 
* Ability to work independently, proactive personality and highly motivated 

Professor:Colin Jones

Type of Project: Master in Industry

Industrial contacts 

Andrea Pedrioli Simulation Engineer, [email protected]

Aldo Cattano Chief Technology Officer, [email protected] 
Skypull SA, via alla Stampa 49, 6967 Lugano

Literature 

* Loyd M. L., “Crosswind kite power”, Journal of Energy 4(3) 106-111 (1980). 
* Kathryn E. Johnson , “Adaptive Torque Control of Variable Speed Wind Turbines”

The energy systems group at CSEMis developing innovative technologies at the intersection of power and energy, control, and data science. In the march towards a fully- decarbonised energy system, we are developing control solutions to make buildings and grids more energy-efficient and more flexible. The current state of the art is model- predictive control to optimize the operating conditions of distributed energy resources. However its practical deployment is hindered by the time and effort required to build models of these resources, and by the inability of most models to reflect the impact of occupants on the energy performance of buildings.

In this context, we are offering an internship on the development of artificial-intelligence algorithms. The objective of the project is to build on state-of-the-art algorithms in deep reinforcement learning to create robust and optimal control strategies. Such algorithms include deep Q-networks (DQN, first demonstrated by Deepmindin 2013 to play Atari games), deep deterministic policy gradient (DDPG, first introduced by Google Deepmindin 2015 for simple motion control tasks), and proximal policy approximation (PPO, introduced by Open AI in 2017 for humanoid running and steering).

The physical nature of the control problems in energy management raises some specific challenges for these algorithms, especially in terms of convergence. The project will therefore combine theoretical aspects, with an opportunity to improve some of the most advanced general-purpose artificial intelligence algorithms, and practical aspects involving the rapid prototyping of software architectures and the implementation of algorithms.

While the objective of the work is clearly industrial, its nature means there will be scope to present the results in scientific journals or at conferences. The student will be part of a team of twelve experienced engineers and researchers.

Professor:Colin Jones

Type of Project: Master in Industry

Contacts: Pierre-Jean Alet, Baptiste Schubnel

Autonomous driving technologies that aim for more safety, comfort, and environmentally friendly vehicles, have been growing rapidly in automotive industry recently. These systems go beyond vehicle-centric assistant applications by integrating traffic and environment information. One of the main challenges in autonomous driving development is to design and then validate the planning and control algorithms in a closed-loop fashion, where both vehicle dynamics characteristics and a wide variety of traffic scenarios are taken into account. The designs should also guarantee optimal performance toward precise tracking, and time/fuel optimality with respect to both vehicle and traffic constraints (i.e. avoid pedestrians and other cars).

One of the main innovation objectives of Siemens PLM is to accelerate the research and development activities of ADAS (advanced driver assistance systems) and autonomous driving technologies. We are continuously looking for outstanding students who are eager to do their Master thesis or internship on a challenging research project in a highly dynamic and international research environment. We have a variety of possible projects available that cover different aspects of planning and control algorithms, ranging from very theoretical to practical. The company also provides various tools to support the research activities, for example, Imagine.Lab Amesim for vehicle dynamics modelling, PreScan for sensor (camera, lidar, radar,…) and traffic environment modelling, and a miniature race car setup for embedded control implementation.

Examples of possible (but not limited to) topics: optimal control and learning control for an ADAS application (i.e. roundabout crossing, valet parking), fast embedded optimal control algorithm, optimal path/motion planning, simultaneous localization and mapping (SLAM) in virtual environment, vehicle dynamics, traffic planning and verification…

Background: control systems, robotics, computer science, or mathematics, familiar with programming. Experience with optimization, vehicle dynamics, ROS or autonomous vehicles is a plus.

Professor: Colin Jones

Type of project: Master

Contact: Colin Jones or Dr. Son Tong

Image result for siemens logo The work will be performed in collaboration with Siemens PLM Software in Leuven, Belgium.

Houses are commonly equipped with PV panels, electric batteries, variable power heat pump and electric vehicles. This means single homes are becoming complex energy hubs which require an efficient energy management system (EMS) that can be configured in a simple way while satisfying complex needs such as improving self-consumption and reducing cost.

The energy systems group at CSEM is actively developing a next-generation energy management system that is based on Model Predictive Control and which can be easily reconfigured through an automated controller synthesis process. A first internship successfully contributed to this project with work focused on forecasting. In this context, we are offering an internship to improve the operation of this EMS by extending its functionalities, focusing on two aspects:

  • The refinement of the predictive control algorithm with specification of complex objective such as peak shaving in the EMS algorithm, and the scale-up to multi-home systems.
  • The buildup of a simulation environment that allows to testbench the controller. The simulation will rely on the versatile physical simulation environment Modelica, for which a collection of house models has already been developed. It will be used to assess the controller performance over long periods of time.

While the objective of the work is clearly industrial, there will be scope to present the results in scientific journals or at conferences. The student will be part of a team of twelve experienced engineers and researchers, and have the opportunity to use real data from test sites and face actual practical challenges of software deployment for energy applications.

Requirements:

  • A strong interest for energy systems and energy management problematics.
  • Good programming experience, preferably with exposure to Python; and an interest in good coding practices (object-oriented programming, unit testing, version tracking with Git).
  • Familiarity with MPC or numerical optimization preferred, knowledge of Kalman filtering and/or machine learning a plus.

Type of project: Master in industry or internship, minimum 4 months.

Location: Neuchatel

Contact: Tomasz Gorecki, [email protected]

Professor: Colin Jones, [email protected]


Past projects (not yet ready)

Projects 2018

Projects 2017

Projects 2016

Projects 2015

Projects 2014

Projects 2013

Projects 2012