Preparing Industry-Ready Engineers: Control Systems for Electric Vehicles

Hello, and welcome to this webinar. My name is George Amarantidis. And for the past three years, I've been working as a customer success engineer at the education team here at The MathWorks. My educational background is in control systems, and specifically focusing on electrification of aircraft. But since finishing, I've also worked in a medical device company and in the semiconductor industry as well.

Our work at the education team is multifaceted. And I'll tell you more about what sort of activities we get into towards the end of this webinar. But one thing we focus on is to try and align university teaching with industry demands. And something that is of particular interest to me is providing an environment for students to acquire hands-on experience with control systems.

In this presentation, we'll go over two main points. Firstly, what are the skills required for an engineer working in the field of electric vehicles? And there might be a slightly biased hint in the title. And second, how can we teach those skills to our university students? But before we start talking about electric vehicles and control systems, let's first talk about the engineers. I have a couple of stories to mention.

So first is the story of Jessica Britt, whose journey really exemplifies the transformative impact of hands-on learning and engineering and more specifically, controls education. So Jessica attended Georgia Tech from 2014 to 2018. And she was heavily involved in student competitions, and specifically the EcoCAR 3.

And now she works at the Argonne National Laboratory. And she credits her success to her practical experience. Initially, she worked at EcoCAR as she hoped it would assist her in securing internships. But her participation really ignited her passion for the field of controls. I feel this is a common trope in students who engage in controls-- at least it definitely was for me-- that until they do hands on activities, they don't fully internalize why this discipline is so exciting.

So back to Jessica. So she held a variety of roles in her EcoCAR team. But for us, the most important ones are in the electrical team, which developed low voltage wiring plans and added control modules and conducted conductivity tests. And she also worked as a modeling simulation controls team lead, where she led the development of Simulink vehicle models, state flow controls code, test plans. And she used dSPACE for hardware-in-the-loop testing.

Jessica credits hands-on projects for defining her career, helping her discover her passion for control systems, and guiding her towards doing further study in the field of controls. Her key takeaway for young engineers is to engage in experiential learning projects. And in my opinion, these projects are crucial for applying classroom knowledge, for refining technical skills, for solving real world problems, and for discovering our engineering passions.

The second story I want to talk about is regarding Graham Iles He began his path to engineering at the University of Bath, here in the UK, where he studied automotive engineering. During his studies, Graham participated in the Formula Student competition, recognized the chance to apply his classroom knowledge in a real world setting.

As powertrain manager on this Formula Student team, Graham was responsible for overseeing the design and implementation of the vehicle's powertrain. During this time, he utilized MATLAB and Simulink, tools that were integral in his academic coursework for simulation and control design problems in a real world setting.

The practical experience that Graham gained in the competition proved invaluable when he entered the job market. His proficiency in Simulink, in particular, was beneficial in securing a position at Mercedes-AMG High Performance Powertrains, where large software models developed in Simulink were commonplace.

Graham's hands-on experience really set him apart during his job interviews. He understood that the key differentiator was not merely the participation in the competition, but also the specific technical and personal skills he developed through it. This approach helped him tackle technical issues effectively in his professional career.

Now Graham is working as a systems engineering team leader at Mercedes. And his whole team uses MATLAB and Simulink for their daily engineering tasks. So now, let's look at the skills that are needed for electric vehicles engineers and control engineers and how multi-domain knowledge is crucial in this field.

So when you think of electric vehicles, what do you first think of? So your current work may be in electrification. So perhaps, you think of power electronics, motor control, and batteries. Or if your focus is on the mechanical side, you may think of vehicle dynamics, or if you're more interested in sensor fusion, maybe automated driving and advanced driver assistance systems may come to mind.

All of these different fields have their own unique challenges. And they're all very complex systems, and there is a lot of modeling and testing required. So you can use MATLAB for modeling those dynamic systems, for defining closed-loop controls, for deploying code to hardware, and even for managing requirements and assisting with validation and verification processes.

So this diagram for the barriers is representative of traditional, if you like, engineering processes. And these barriers are very often replicated in how engineering is being taught in universities. An integral part of successful engineering firms is a process we call model-based design. And it's crucial for students who want to work, not only in electric vehicles, but in any kind of engineering systems, really.

So instead of siloed mechanical, electrical, and control systems, you can optimize your design in a single simulation environment. You can detect errors and mismatch with your requirements by having feedback between the requirements and the system level design with continuous verification. You can automatically generate embedded code ready to deploy on your microcontrollers.

You can reduce cost by running hardware-in-the-loop simulations. And there is constant testing and verification as feedback loops in all of these different stages. There are many examples of companies using model-based design. But the recent one I really like is with Bosch, where they developed a control system for their eBike within a tight schedule.

Their team had just nine months to design and produce customer samples for the drive system, and just five more months before the system was introduced at the EUROBIKE Trade Fair. So model-based design was instrumental in enabling them to deliver the embedded eBike drive system controller within the deadline.

But it's also really important to note that they didn't just use MATLAB and Simulink for their prototyping, they also generated production code that met safety standards. So industry leaders across various domains such as automotive, aerospace, industrial automation, and so on, rely on MATLAB and Simulink for their control systems design and development.

Companies like Nissan, ABB, Jaguar, Land Rover, Toyota, NASA are looking to hire people with MATLAB and Simulink knowledge. But it's not only the industry leaders, but also many startups that use our tools, such as a swift development time and a low time to market is crucial for them in order to be successful.

So why do people use MATLAB and Simulink? I have a bunch of quotes here. I'm not expecting you to read any of them. But just to highlight the key points, MATLAB and Simulink helps them work well with others. It extends their capabilities. They can scale their applications. They save money, and they save time.

So the main ideas that keep cropping up is that MATLAB and Simulink really accelerates these projects. Our tools enable non-specialists to branch out. They promote collaboration, and they reduce development time to code generation. And as you'll see in a bit, these skills are invaluable to recent graduates for securing roles in the industry.

So talking about the industry, what does a control engineer or an EV engineer do in the industry? So let's take a step back and look at some job postings relevant to EVs. So I found these on LinkedIn, specifically. So after studying at a university, what jobs do these students apply for in the industry? I've got three example roles.

The first one is from a multinational, originally French consulting company who works with most large automotive companies. They're looking for a mechatronics engineer to be part of their ADAS, an autonomous vehicle team. Specifically, they're looking for an engineer who has understanding of modeling and parameter identification.

They have experience with modern and advanced control, such as state-space and MPC. And of course, they have knowledge of ADAS. The second job is from a large EV manufacturer in the US, and they are looking for a vehicle motion controls engineer who specifically has knowledge of vehicle dynamics, knowledge of multi-domain systems, of optimal control. And they're also needed to be well versed in MATLAB and Simulink.

And lastly, the battery management systems role is with a British automotive company famous for their sports cars. The engineer is to actively support the specification, design, engineering, and delivery of vehicle control systems, including battery management systems within the hybrid and EV systems teams. They'll also be interacting with the powertrain team.

The desired candidate needs to be familiar with embedded control and hardware-in-the-loop simulation as well. They'll also be working quite closely on functional safety with ISO 26262. So if we were to summarize these in a bit, what is the industry looking for in the recent graduates?

Well, the industry's looking for control engineers with diverse skill sets beyond basic control theory. Our graduates need to be familiar with more advanced techniques such as optimal control and MPC. They need to be familiar with MATLAB, and Simulink, and model-based design. They need to have an in-depth understanding of their application areas.

And they need to be experienced with hands on projects and hardware-in-the-loop deployment. So here is a map of feedback control systems created by Brian Douglas, whom you may know from his popular Tech Talks videos on control systems.

So when I learned control in university, a few years back now, there was a running joke that in industry all people use is PID control. And granted, PID control is really powerful and does solve a lot of problems in the automotive and electric vehicle world. But with problems becoming increasingly more complex, PID just doesn't cut it every time.

For example, for ADAS, model predictive control is very commonly used in the industry. For Active Disturbance Rejection Control, or ADRC, it offers high performance solutions for motor control as it allows for handling non-linearities and provides better performance compared to traditional PID.

Extremum seeking control is great for anti-lock braking systems, as it continuously and adaptively optimizes braking performance in real time, ensuring maximum tire road friction and improved safety. And it's not only the industry that requires those skill sets. There are many applied controls PhDs and postdocs that require very similar skill sets.

And I found these ones, again, online. I'm sure there are plenty more like this. And it's not just about understanding controls or having knowledge of mechanical, electrical engineering, and so on. But we need to have an understanding for the surrounding fields and the interplay among them, as well as being familiar with hands on projects. Very identical skills, if you like, between the industrial job openings and the academic job openings.

So control systems are at the heart of countless applications across various domains such as mechanical, electrical, aerospace, chemical, and others. So engineers working control systems must have a broad understanding that spans multiple engineering disciplines.

So on an electric vehicle, mechanical engineers might be concerned with the physical systems and its controls, while electrical engineers may care more about the circuitry and power transmission. Computer engineers might want to use algorithms for fault detection in ADAS. And these complex systems require a multi-domain approach to be successful. We can't just have a siloed viewpoint of each individual discipline.

So for multidomain systems to effectively operate and to link between those domains is when control engineers are able to see the bigger picture. MATLAB and Simulink ecosystems enables multi-domain curriculum in control systems and equips students with a versatile skill set that is needed to innovate in a rapidly evolving industry such as electric vehicles.

So linking back to the roles we looked at, both in industry and in academia, students need to have the following skills. First and foremost, interdisciplinary foundations-- model engineering systems such as autonomous vehicles, smart grids, robotics, manufacturing.

Integrate mechanical, electrical, and computational components. A control system curriculum that embraces a multi-domain approach, equips students with foundational knowledge to understand and bridge these interconnected disciplines is paramount.

Secondly, systems thinking, a multi-domain curriculum fosters systems thinking, where students learn to view control systems, not just as an isolated entity or applied mathematics, but as a part of a larger integrated whole. And this holistic perspective is crucial for designing and analyzing robust systems.

And lastly, or thirdly, real-world experience-- so hands-on projects are crucial for a multi-domain curriculum, as they provide practical, real-world experience. Students may work on industry-relevant projects that require multi-domain knowledge, making them more prepared for similar challenges in their careers. OK, so all that's great. But as educators, how do we achieve that? How do we get started?

So initially, we had the look at what our students need. So now let's change gears and talk a bit more about how can we prepare the students for the industry, and also how do we keep them engaged while we do that? So we'll look into three different things-- project-based learning, interactive tools, and flipped classroom resources to help you and support you in the teaching of the future engineers.

So if you attended any controls conference in the summer, you probably ran into me or some of my colleagues, asking you questions and doing questionnaires about how do you teach control systems and what your needs are. So yeah, recently we conducted these surveys in different conferences, where we asked academics such as yourselves about the challenges that you face. And here are some of the key themes that kept coming up.

So some of the ideas are keeping students engaged during the classroom and finding intuitive ways to teach complex control topics. Introducing students to hands-on projects to showcase practical applications and increase motivation has always been challenging, and this year, even more so.

If you all mentioned that they want to highlight the more applied problems of control systems such as deploying to hardware and testing-- great stuff, all very useful skills that are needed in industry and academia. So how do we do it? How do we go from the left hand side of our Ogata textbook to the right hand side of the applied control systems?

So as Jessica and Graham Iles mentioned in their interview, what really speaks to students is interesting hands-on projects. So on the right here, I've got a photo from TEDI-London. So TEDI-London is a newish university in the UK where teaching revolves around flipped classroom and project-based learning.

As a matter of fact, they've just had their first cohort of graduates this year. So huge congratulations to them for this milestone. So on this photo, students actively are engaging in project-based learning. They all appear interested and engaged.

And in this case, they're taking advantage of the hardware support packages they provide to deploy code that they've developed in Simulink onto hardware. And the hardware of choice for them is Arduino with a Zoomer robot. Furthermore, a colleague of mine, Andreas Apostolatos was heavily involved in supporting TEDI in designing this course.

So this is an example of a project that we have actively fostered and supported from MathWorks. There are other successful courses that do PBL. For example, in King's College London, which is one of the oldest and most renowned universities here in the UK, they use Simulink, Simscape, and Arduino to allow students to experience an end-to-end workflow for industrial design.

By implementing a digital twin, learners can validate their modeling and control systems by comparing the results between simulation and physical prototype. And these are just some examples of how colleges and universities have brought their control curriculum to life. In order to adequately prepare their students, they've created engaging hands-on materials.

So if you are creating such interactive materials, or virtual labs, or project-based learning projects, or anything innovative really that you're using to teach fundamentals of control, or advanced controls, or model control, I would love to hear more about it and potentially collaborate. So please do reach out.

So although we'd all love to use more project-based learning in our classrooms, it may not be always easy to achieve due to other limitations, such as classroom size or specific rigidity of the course that we teach. But how can we still engage our students and ensure that they're adequately prepared for impactful careers?

Well, in general, we want to enhance our traditional teacher-centric model and promote intuitive understanding through hands-on exploration. A great way to do this is using interactive tools and apps. For example, some great work on that has been done by Anthony Rossiter by creating a variety of different apps to highlight fundamentals of control.

He's from the University of Sheffield. And this project was called Control 101 toolbox. And you can find more information on that on File Exchange. We also wanted to use real applications and examples where the students get to implement what they've learned in the classroom. And the sooner we do that, the more likely they are to stay engaged and learn more about their engineering interests.

So we can't keep telling the students, keep learning the mathematics, and it will all make sense someday. They want to see what's the impact of what they're learning and how they can change the world around them. One way to enable engaging and interactive learning of controls concepts is through apps and virtual labs.

So these apps, we have a variety of them on File Exchange. And they're either created by other academics like yourselves or by MathWorks. You can also create your own through the MATLAB app designer. Furthermore, we have several other virtual mechanisms or digital twins, if you like, that you can use and adapt for your courses. The one I tend to use is the inverted pendulum.

Whenever I do guest lectures, it's one of my favorite examples to bring up and do some embedded control with. Great coursework that I've seen developed as well, is taking the cruise control example and extending it to adaptive cruise control of a platoon. And this was for a-- if I recall it correctly-- second year controls coursework. So this was the first time students had really interacted with controls disciplines.

But you don't need to make full apps like Anthony's Control 101 to make materials more engaging. Simply by converting your m-files to interactive notebooks and adding a couple of UI elements can really boost interaction with the content. You're just highlighting to your students, hey, you can tune this parameter and see what impact that has to the response of your system and so on.

And you don't just have them explore blindly the script. And this can really help them with feeling less pressure with regarding interacting with code. And let's not forget about Simulink. Control systems are often taught using block diagrams, illustrating the relationship between different system components.

And Simulink, of course, not only provides a block diagram environment, letting students interactively design and implement control systems and physical models in the exact same way they learn them in textbooks, but it also provides a full ecosystem that allows students to go from initial concept, and mapping to requirements, to model deployment of the hardware.

We also have a variety of resources that can support your flipped classroom. By encouraging students to use these, instructors can devote more time in class for problem solving, hands-on activities, and interacting with their students on a personal level, rather than spending time on tool or concept familiarization.

Online content also means high quality educational material is available to students across institutions. Great examples of these are our Tech Talks, where educational videos explaining complex controls and engineering concepts are becoming more accessible. So we have a large library of these Tech Talks, covering topics ranging from PID control and system identification, all the way to reinforcement learning.

We also have self-paced courses called Onramps, which helps students get started with MATLAB, Simulink, and control design. In addition to broaden access, online materials and even the Tech Talks as well, actually, are very useful for industry professionals who seek to improve their skills on their own time.

So there's a variety of excellent content across disciplines, such as image and signal processing, electronics, deep learning, and others. This is a great way to be introduced to new topics. So currently, I am doing an Onramp on generating C code from MATLAB.

So all these resources we discussed today and many more are available on our course website, including curriculum, virtual labs, apps, projects, videos, and more. So make sure you visit the site where you'll find resources, not just on control systems, but other related disciplines as well.

A few last notes from MathWorks. We also support dozens of student competitions worldwide each year in all kinds of fields, including automotive, aerospace, and robotics, to allow the students to get hands-on understanding of the skills they learn in university.

We also have the Student Ambassador program where MathWorks sponsors students, coach their colleagues in MATLAB and Simulink. They interact with their peers, and they arrange workshops, hackathons and other fun activities for the community. And lastly, we've got customer success engineers like myself.

As I mentioned in the beginning of the presentation, we consult with faculty, with researchers, and executives to support their initiatives. Oftentimes, collaborations may look like curriculum development, and updating existing curricula to meet accreditation requirements, or save the instructor time, and it improves engagement.

Other times, we might do guest lectures and help the students understand in a more hands-on way what's being taught in the classroom already. Another thing we do is we contribute to industrial advisory boards and trying to align industry needs with what is being taught in the university. We also advise on the latest technology for multidisciplinary research. And we support research groups in their commercialization efforts.

And we also support student communities through competitions and project-based learning. As I said earlier, as a call for action, reach out to us and tell us more about your research needs. If you're developing materials, or if you're thinking of developing materials that may be using MATLAB and Simulink for electric vehicles or for controls, reach out to us. Tell us more about it.

And also, here are some other resources on control systems materials, electric vehicles, self-paced courses, and the Tech Talks I mentioned earlier. Thank you very much for your attention. And I'll now take any questions that you might have. Thank you very much.