Research Opportunities at the UIUC Engineering System Design Laboratory
This public Google doc is linked to from the ESDL research home page:
Postdoctoral Research Opportunities
Graduate Research Opportunities
Undergraduate Research Opportunities
Foundational Coursework and Technical Background for ESDL Research
Background Courses for Co-Design
Courses in Mechanical/Physical System Modeling and Design
Additional Courses Relevant to Co-Design
Background Courses for System Architecture and Topology Design
Preparing for an Independent Study Project
Identifying a Graduate Student Mentor
Writing a Research Project Proposal
Specific Instructions for Graduate Students
Notes for Students Applying from Institutions Beyond UIUC
At the moment (updated 6/29/2021) there are no ESDL postdoctoral research fellow openings. We are actively pursuing new funded research projects, so new openings may emerge in the future. Typically we need candidates who have a strong background in numerical methods and computing, engineering design, and expertise in at least one of the following: time-domain controls (ideally also direct optimal control methods), and computational mechanics relevant to ESDL applications (e.g., wind turbines [aeroelasticity, structural dynamics, hydrodynamics, generator systems], hydro turbines, intelligent structures for spacecraft attitude control [structural dynamics, rigid-body dynamics, piezoelectric materials, spacecraft dynamics], and electro-thermal systems [computational heat transfer, conjugate heat transfer, electromagnetics, cooling system fluid dynamics]).
All recent PhD student openings have been filled.
Non-thesis UIUC MS students who need to complete an independent study project can review the available projects listed below for undergraduate research. A similar process needs to be followed, including identifying a senior graduate student mentor, and writing a short research proposal.
In addition to the projects described below, undergraduate students are encouraged to learn about ongoing research projects being conducted by graduate students. There may be opportunities to work with graduate students on their projects beyond what is listed here. An important prerequisite for beginning research within the ESDL as an undergraduate student is to identify a current ESDL graduate student who agrees to mentor you. Please also be sure to read the section below outlining important coursework and technical background for ESDL research projects. Please also review the section below that provides guidelines for students preparing for independent study projects. Many of the same guidelines apply to students interested in paid opportunities.
ISE Undergraduate Students: Since May 2014, the ISE department has been offering support to pay undergraduate researchers to work on projects like those listed below. If you are an ISE student, you may be eligible to submit a proposal to work on one of these projects. Obtaining an ISE undergraduate research scholarship is an excellent way to have an opportunity to get paid to work on a research project. Funding from research grants most often only go to UIUC undergraduate students who have demonstrated skills that are relevant to funded projects.
Non-UIUC Undergraduate Students: We have a large number of highly-qualified undergraduate students at UIUC who have the opportunity to demonstrate their work ethic, technical skill, and productivity through initial volunteer research opportunities. In most cases funded undergraduate research positions (and often opportunities for academic credit) go to UIUC students who have already proven themselves through these initial research activities. If you are a non-UIUC student seeking a funded position, if you do not have your own external funding it is unlikely that you will secure a funded research position within the ESDL. If you are a non-UIUC student and do have your own funding, you may be considered for an ESDL position if you: 1) have very strong academic/technical credentials, 2) have strong recommendations from experts in your research area, and 3) have strong written and oral English communication skills.
Note for non-thesis graduate students: Some of these projects may be appropriate for independent study projects to be performed by non-thesis graduate students. You will need to identify a graduate student mentor (more senior thesis graduate student), and write a short research proposal as described in the section below on preparing for independent study projects.
Undergraduate Research Position: Educational Module for Quantitative Design Coupling Analysis
Position Type: Paid or academic credit. A paid appointment is only available for US citizens, except for students who are successful in obtaining their own independent undergraduate research funding (such as the ISE undergraduate research scholarship). It is expected that this project will require at least two semesters of effort. Please contact Prof. Allison directly for more information.
Project Background: Many engineering systems involve two or more sets of design decisions, such as mechanical system design and control system design decisions. If particular decisions made within one set of design variables impacts how decisions should be made for other design variable sets, we say that these design decisions are coupled. Many engineering design problems can be formulated as mathematical optimization problems. In this case, design coupling can be quantified using derivatives between different sets of design problems. Analysis of these derivatives can provide valuable insights into the nature of particular design problems.
Project Objectives and Outcomes: The objective of this project is to create learning materials and an interactive learning tool based on one or more engineering design optimization problems to aid individuals in learning design coupling concepts and analysis methods. This education module will be designed to be useful for: 1) senior undergraduate engineering students, 2) practicing engineers, and 3) engineering researchers hoping to apply design coupling analysis to system design challenges. It is anticipated that the learning module will include interactive tutorials based on MATLAB.
Student Qualifications: The ideal student qualifications include completion of SE 413 (Engineering Design Optimization), excellent writing skills, proficient in MATLAB programming, background in general numerical methods.
Other ESDL undergraduate research opportunities: Please check directly with Prof. Allison or an ESDL graduate student working in your area of interest for possible undergraduate research project opportunities.
Often students ask what types of courses and technical background are relevant to ESDL research projects. This section summarizes the types of courses that provide helpful background for students wanting to do research with the ESDL. Most of the courses described here are senior or graduate level courses. While these courses are open only to upper division or graduate students, there are still opportunities for lower-division undergraduate students to engage in research with the ESDL. It may be helpful for lower-division undergraduate students to make sure they do well in courses that are prerequisite to the senior and graduate-level courses described below, and also to be willing/able to learn advanced material independently as required for research projects.
Most ESDL research pertains either to co-design (integrated design of dynamic systems, including actively controlled and autonomous systems), or system architecture/topology design (primarily using generative design algorithms). Some projects involve physical prototyping/experimentation/validation, so lab experience and other hands-on background can be very useful, especially with mechanical or mechatronic systems and fabrication of these systems. We also seek to learn how co-design and other advanced design methodologies may be utilized by existing design organizations, so understanding of industry design processes is also important (e.g., AE 542, industry experience) .
We are interested in engineering design applications with potential for profound impact on humanity, such as energy sustainability (e.g., wind and wave energy, hybrid powertrains for agricultural vehicles, efficient power electronics) and the creation and advancement of tools for scientific discovery (e.g., new methods for space telescope design). ESDL research does not emphasize the study of consumer products.
IMPORTANT: Many students inquiring about research opportunities in the ESDL have interest/background in mathematical optimization or other topics in operations research, but have little background in physics or design of physical engineering systems. Nearly every research project within the ESDL is connected strongly to physics-based design (i.e., design decisions regarding physical systems that are based on analysis using physics-based models) and practical physical engineering applications. While we need students who are strong in optimization and algorithms in general, we need students who also have good background in physics-based analysis of engineering systems, and intuition for how physical engineering systems work and should be designed. The level of this expectation changes with the level of student. Lower-division undergraduates should at least have a strong interest/enthusiasm for these topics, and they can develop knowledge in these areas as they engage in research projects and take more courses. Candidate graduate students, on the other hand, should already have some strong foundational background in the design of physical engineering systems and physics-based analysis. In addition to expertise in physics-based analysis, experience with physical implementation of engineering systems is especially helpful for developing the intuition required to perform research in physics-based engineering design.
It is generally much easier to teach the required mathematics and numerical methods to an individual who already has a good understanding of and intuition for engineering design and applications than it is to help a student with a strong mathematics and computer science background gain the requisite knowledge of engineering design. Except for smaller, contained projects where the objectives require specific, limited knowledge that excludes physics/engineering background, we generally do not accept students into the ESDL who only have mathematics/computer science background and have no experience with relevant engineering applications (e.g., mechatronic/mechanical/structural/power electronics systems).
Co-design research incorporates knowledge from several areas, including: control system design, design optimization, and modeling/design of physical systems. Students should have background in one or more of these areas (ideally all three). It is also recommended that you read ESDL journal articles 9 and 10 to gain background in co-design if this is your area of interest.
Most of our work involves time-domain methods (modern control/state-space control), so having taken a course in state space controls is foundational for this work, such as:
or other similar courses. Graduate students working on co-design need to have taken at least one course in state space control systems (i.e., one of the above courses or an equivalent).
Most co-design work also involve some form of optimal control. Because of the need to manage inequality constraints pertaining to physical systems (and other reasons), we primarily use direct optimal control methods, such as direct transcription or pseudo-spectral methods. Graduate students working in the area of co-design should plan on taking at least one course in optimal control relatively early during their graduate program. Options at UIUC for optimal control include:
Other courses in the area of control systems that may be relevant, depending on research topic, include:
Other topics of interest to the ESDL in the area of control systems include Reduced Order Modeling/Space Mapping, and design and control of autonomous systems.
Feedback control systems are the first layer of intelligence in autonomous systems (e.g., tracking a set point). Higher levels of intelligence allow engineering systems to operate more autonomously. Learning how autonomous systems should be designed differently from passive or human-operated systems is a core research interest of the ESDL. Strategic co-design studies and analysis of resulting designs can help reveal more fundamental principles of Design for Autonomy. Solving the co-design problem for autonomous systems remains an open research question, and will require incorporation of additional layers of intelligence into integrated co-design methods (including supervisory control, planning, and decision-making systems). Students interested in this topic should take courses in Artificial Intelligence and Machine Learning, Decision Theory, and other related topics.
While Linear Programming (optimization of linear objectives with linear constraints) is relevant to some of our work (e.g., Model Predictive Control), in most cases we are interested in solving Nonlinear Programs (optimization problems with nonlinear objective(s) and constraints). This is due primarily to the nature of design constraints for physical systems. ESDL graduate students should develop a strong foundation in nonlinear programming, especially as it applies to design optimization. In co-design we seek to optimize simultaneously both control and physical system design. The most successful co-design methods so far are based on nonlinear programming.
The following courses cover nonlinear programming. Some address the connection between mathematical optimization and engineering design (this connection is essential for ESDL researchers to understand well). Graduate students should take at least one course in nonlinear programming that is mathematically intensive, and one that covers engineering design optimization.
In addition to gradient-based methods and optimality conditions, students should also learn derivative-free methods, including those with alternative optimality conditions (such as pattern search), and heuristic methods (such as Genetic Algorithms) without optimality conditions. The variety of problems solved in ESDL research requires versatility in our optimization toolset. Other optimization topics that are important to our work include surrogate modeling, multi-objective optimization, and multidisciplinary design optimization (MDO). An introduction to many of these additional topics is provided in GE 413. IE 513 covers MDO.
One of the most important distinguishing characteristics of ESDL research is our unique approach to co-design. Most other co-design researchers take a strongly controls-centric approach to their work. This has led to fantastic advancements in co-design, but has also resulted in missed discoveries and methods that do not account fully for the complexities of physical system design. These issues are detailed in ESDL journal articles 9 and 10. We seek to take a more balanced approach to co-design where physical system design considerations are treated in a comprehensive manner. We seek to make discoveries at the interface between physical and control system design.
This type of research requires students who have a strong foundation not only in controls, optimization, and algorithms, but also in the modeling and design of physical systems. At the ESDL we aim to help students become experts at interfaces, including the interface physical and control system design.
ESDL students need to learn how to think as strong design engineers. Developing this capability requires a combination of strong analytical knowledge and intuition for the physical systems being designed. ESDL students should have a passion for engineering design and for impacting the world through better engineering design. Developing passion and intuition is aided by hands-on experience with engineering applications. In addition to modeling, optimization, and design work, we engage in some physical prototyping and testing. Both undergraduate and graduate students have opportunities to get involved with the hands-on aspects of ESDL research.
Undergraduate students can develop expertise in physical system modeling and design, as well as design intuition, through their engagement with ESDL research projects. Candidate graduate students should already be design-minded engineers with solid background in some aspect of physics-based modeling of engineering systems.
The design applications we study often involve mechanical or structural systems, so background in solid mechanics, finite element analysis, and mechanisms is normally very useful. We are often interested in the dynamics of these systems, so background in structural dynamics and multi-body dynamics can be important depending on the research project. We are engaged in projects that involve a variety of other analysis disciplines, including heat transfer, fluid dynamics (including rheologically complex fluids), and electromagnetics. Many of our studies involve the coupling between multiple analysis disciplines (e.g., aeroelasticity captures the coupling between structural and fluid dynamics for structures affected by fluid flow). These types of analyses are known as either multiphysics or multidisciplinary analysis. MDO methods seek to generate system-optimal solutions that account for coupling between different disciplines, helping to improve system performance compared to more conventional design techniques.
Undergraduate students should seek to excel in physics-based analysis courses, and seek to understand how to model physical systems in a way that is useful for supporting design decisions. Upper-division undergraduate students can take courses as electives that help to strengthen their design and modeling expertise. Some of these courses listed below are appropriate for upper-division undergraduate students. Graduate students should seek to develop a rigorous foundation in at least one physics-based modeling area that is relevant to their research project, and seek to develop strong engineering design intuition for the applications they are working on.
Design Courses: In addition to the design optimization courses listed above, the following courses should be considered:
Courses in Engineering Mechanics, Dynamics, and Computational Methods: While the specific courses a student should take is largely dependent upon research topic, here is a sampling of fundamental courses in mechanics that are likely to be useful in ESDL research. ME and TAM offer many additional courses in this area (including many advanced courses) not listed here that may be useful for graduate students. Please seek guidance on selecting appropriate courses for your research topic. Graduate students should work to develop significant depth of expertise in physics-based modeling in the analysis domain(s) relevant to their research topic, with particular emphasis on developing models that are useful for design and design optimization.
Courses in other analysis domains, such as fluid mechanics or heat transfer, are not listed above but are important for specific research topics within the ESDL.
Additional Courses Relevant to Modeling and Design:
Students may need to take more integrative or application specific courses, such as some of those listed below:
System architecture here refers to the set of components that comprise a system and the relationship between these components. Much of the work in co-design described above assumes that system architecture is predefined, and the objective in co-design is most often to optimize the design and control of components in the system to maximize system utility without changing system architecture. System architecture often can be represented using networks (or graphs) where nodes/vertices represent discrete system components, and edges between nodes indicate the presence of a relationship between components.
Differences in system architecture design are topological in nature, so in a sense system architecture design involves topology optimization. The topic of topology optimization, however, is a rich area of research that focuses on the physical design of systems governed by continuum mechanics, such as structures. While some work in topology optimization involves multiphysics and complex anisotropic materials, most work in topology optimization focus on design problems that are homogeneous in nature. For example, in structural topology design, the objective is to determine where in a design domain it is most advantageous to put material, but every bit of material behaves in the same (or similar) way. System architecture design is a broader topic in that we are making discrete decisions about what components to include in a design, and the functionality of different components often is fundamentally distinct (e.g., in powertrain design an engine serves a completely different role from a gearset).
System architecture design and topology optimization are similar topics, but system architecture design requires special techniques for handling the heterogeneity of system components. For example, in structural topology optimization we can represent a particular design using an undirected graph where every node represents functionally identical components (i.e., steel elements that resist force in a predictable way), whereas in hybrid powertrain architecture design each node in a graph representation corresponds to a unique component. The discrete differences between components can be represented using graph node labels or “colors”.
At the ESDL we are interested in investigating new design methodologies for both system architecture and topology optimization. In both cases, methodologies must explore a range of different candidate design topologies, and perform a quantitative comparison between candidate designs to work toward identifying an optimal (or approximately optimal) design. The three topics of exploration, quantitative comparison, and optimization are discussed below.
Exploration/Representation: Exploring system architecture candidates requires that we have efficient means for representing network structures, and for navigating the space of system architecture design candidates. Direct network representation methods (e.g., dealing with edge lists or adjacency matrices directly, as is the case with ground structure methods--see here as well) are straightforward conceptually, but have many disadvantages (difficulty scaling up, difficulty in application to more complicated design problems, require penalties or repair mechanisms). Indirect network representations solve many of these problems, and enable the embedding of intelligence within network representation methods to focus the design search and automatically satisfy specific design requirements. We are interested specifically in generative design algorithms (used traditionally in art and architecture) as an efficient means for representing system architecture. See conference papers [20] and [26], as well as thesis [4], for examples of our work in this area.
Work using generative design algorithms (GDAs) for system architecture and topology optimization requires knowledge across a range of topics. Students need to develop algorithmic creativity to explore, implement, test, and adapt new algorithms for system topology representation. Strong programming skills are essential for work in GDAs. Students also need to understand graph/network theory. Other mathematical topics, such as probability and statistics, may also be important depending on the type of GDAs being investigated. Students interested in more theoretical aspects of this work should have a strong mathematical foundation, including real analysis and theory of computation. While work in GDAs lends well to more abstract research, it still must be grounded in physics-based engineering design optimization. Students conducting research in system architecture design using GDAs still must have or develop a strong foundation in quantitative physics-based engineering design.
Courses important for system topology representation and exploration using GDAs include:
Quantitative Comparison:
(in progress)
Optimization:
In addition to gradient-based methods, students need to develop extensive knowledge of heuristic and other gradient-free methods (GAs, simulated annealing, pattern search, etc.).
Significant overlap in the required background described above in: optimization, modeling, design.
While the following is written primarily to undergraduate students interested in doing independent study research with the ESDL, most statements apply also to graduate students who are interested in completing an independent study project with the ESDL.
If you are interested in academic credit in conjunction with undergraduate research performed in the ESDL, please read the following guidelines. Academic credit is received through registration in an independent study course (for example, GE 497 or IE 497). You can register for between 1 and 4 credit hours when taking an independent study course. The number of credit hours will depend on the scope of the project you define, and will be a decision made between you and Prof. Allison. If you are not an ISE student (IE or GE), you will need to check with your undergraduate office or advisor whether taking an ISE independent study course will count toward your degree. It might be possible for Prof. Allison to advise a non-ISE independent study course, although whether this is a possibility will depend on the policy of your department.
To get started, you need two things: 1) identify a graduate student mentor, and 2) write and have approved a short research proposal.
Undergraduate students working on an independent study project in the ESDL work closely with a graduate student mentor (they meet at least weekly, and in some cases spend several hours per week working together depending on the nature of the project). Often undergraduate research projects are defined to support a larger project that a graduate student is leading. If the project is part of a larger collaborative project, the undergraduate researcher will need to attend these additional collaborative project meetings. Sometimes undergraduate projects involve curriculum development instead of a research project. If you have not done so already, please read through the ESDL website to learn:
Please be proactive in contacting graduate students to learn more about their current or upcoming projects. Graduate student email addresses are listed on the ESDL website. You will need to identify a graduate student mentor in your proposal. A graduate student mentor must be:
In addition, the project needs to be mutually beneficial. That is, in addition to there being an educational benefit to the student completing the independent study project, there must also be a net benefit to the graduate student mentor. For example, one question that should be asked is “Would it be less effort for the graduate student to mentor and train a new student to complete a project than to instead complete the project him/herself?”.
To register for an ISE independent study course you will need to fill out a form available from the ISE undergraduate office. You will need a signature from Prof. Allison, and we will need to decide on the number of credit hours for the project. Before Prof. Allison will sign an independent study form he requires a short research proposal (around 3-4 pages is usually sufficient). The proposal should include:
This proposal should be submitted to Prof. Allison at least one week before the start of the start of the semester you hope to enroll in independent study. He may provide feedback on the proposal that will require adjustments to your plan. You should work closely with an ESDL graduate student mentor in preparing this proposal before submitting it to Prof. Allison.
One way of looking at this proposal is that it serves as a syllabus for the independent study course. It is essential that a problem is identified and that a plan is in place before the semester begins. This may require some significant preliminary work, including reading research articles/theses/textbook sections or other materials as recommended by your potential mentor. We also recognize that with research projects, plans may need to adapt as discoveries are made, so the research plan in the proposal should be viewed as an initial structure that might be adjusted later in the semester in consultation with your mentor and Prof. Allison.
As you are preparing your proposal you should probably meet with one or more candidate graduate mentors several times to get feedback, instruction, and guidance. You should also meet with Prof. Allison at least once before submitting your proposal. Please ask graduate students for small tasks that you could do to get some experience and contribute to ongoing projects. Graduate students can also point you to important references to read as you are learning about a subject and preparing to write a proposal. You should also try reading some of the research publications written by ESDL researchers that are available on the ESDL website. Depending on what courses you have already had, the details may be difficult to understand, but it is still useful to try and get a high-level understanding of the research conducted within the ESDL.
Another helpful thing you can do is attend the ESDL group research meeting, held every Tuesday at 9:00 am* (except holidays, or when Prof. Allison is out of town; summer full group meetings are held only occasionally). This is usually held in TB room 303, although sometimes may be held in rooms 304 or 406. In these meetings we discuss lab business items, and group members take turns presenting updates on their research projects.
*This day/time for ESDL group meetings is accurate as of 02/27/2017. At times the meeting may be cancelled for breaks between semesters or to accommodate conference or other travel by group members. You may want to check with Prof. Allison or an ESDL graduate student whether the meeting will be held on a particular week.
Graduate students who are interested in completing a graduate-level independent study project with Prof. Allison should still identify a thesis graduate student mentor from the ESDL as described above. While preparing for and after beginning a project, you should meet regularly with your mentor and attend any additional research meetings as appropriate. In most cases Prof. Allison will not be able to schedule regular weekly meetings with students enrolled in independent study projects. He meets weekly with thesis graduate students, and holds regular research meetings for collaborative projects. One-on-one meetings with independent study project students are held on an occasional as-needed basis.
Note for non-thesis MS students: These guidelines apply also to ISE non-thesis MS students who are looking for an academic advisor and are interested in working on their MS project under the guidance of Prof. Allison. Please see the above information regarding research project proposals, as well as the section on technical background/coursework that is relevant to ESDL work. Please also be aware that if your offer letter from the ISE department did not include an offer of TA or RA funding, and if you are a non-thesis student, the ESDL generally does not provide RA funding. PhD students have priority for RA appointments, followed by MS thesis students, but often appointment decisions also include important factors such as specific available projects and student background.
Joining the ESDL as a thesis student (MS or PhD) is highly competitive, and if a student joins as a non-thesis MS student, there is no guarantee that he/she will have an opportunity to switch to a thesis degree. Beyond funding, the primary constraint is Prof. Allison’s time. He meets with thesis students weekly, and has a limit to the number of students he can advise at any one time and dedicate sufficient attention to guiding their research.
What is the difference between graduate independent study and an MS thesis?
We often receive requests from students seeking internship or other temporary research opportunities within the ESDL. If this is the case, please read carefully the following notes: