A Good Mechanical Engineering Project
Having a good project to showcase in portfolios and discuss during interviews is essential for any mechanical engineer.
Coming up with a strong project idea and successfully executing it can be quite challenging. All too often, mechanical engineering students either have ambitious project ideas that never get completed (or started for that matter) or choose something that turns out to be more of an "I followed an Instructable" endeavor (like building a classic motorized skateboard).
In this guide, we aim to walk you through our thoughts on how to ideate and execute a successful project, accompanied by an example from a project we completed a couple of years ago.
The Idea
Generating a solid project idea might actually be the most challenging part of the entire process. You need to select something that you can realistically complete within a six-month timeframe (one semester), have the necessary tools to build, and produce results that are measurable and demonstrable in a portfolio and presentation.
Good ideas don't necessarily have to be groundbreaking or entirely unique. We believe that good projects exhibit three key characteristics:
Exhibits Technical Hardware Design: This can encompass a wide range of things and doesn't necessarily require reinventing the wheel. The challenging hardware aspect could involve packaging something relatively simple into a small, aesthetically pleasing form factor that is highly presentable. A prime example of this was a project undertaken by some of our classmates—an automatic chess-playing robot. In reality, getting an X-Y 3D printer-like mechanism to work should not be a challenge for junior or senior mechanical engineers, but packaging it into a normal-sized chessboard that is only a couple of inches tall presents a more difficult constraint.
Incorporates Electrical Components: We live in an era where mechanical engineers are expected to be multidisciplinary. It's not sufficient just to design parts; you need to make them move. Additionally, when it comes to demonstrations, everyone loves something that can actuate! From a strictly mechanical standpoint, manual designs are more technically pure and some of the older generation live and die by this philosophy (looking at you unnamed Professor). However, when it comes to wow factor, it’s hard to beat the magic of automation, and it is guaranteed to make your project more interesting/appealing to someone reviewing it.
Uses Clever Code: As a mechanical engineer, you’ve probably dabbled in Matlab and Python at some point in time. Integrating these skills into your project is crucial (yes, you can find someone else to do the coding, but we recommend doing it yourself). Your electronic components won’t be able to do much without some coding, and a static robot is a sad one. That being said, we don't expect you to produce perfectly streamlined, efficient code. Personally, we subscribe to the philosophy that computers are just rocks we struck with lightning and tricked into pretending to think. A hack is only a hack if people can tell it is. 95 times out of 100, no one is going to look at your code base but you, so comment your code however you want, use 10 if statements in a row, give your variables fun and meaningless names. No one will ever be the wiser [probably]. All we care about is that it functions to an acceptable level (depending on the project requirements, this may be rather more involved than we’ve painted. Either way, no need to go learn C in order to scrape the bottom of the barrel for an extra 50 milliseconds of performance unless necessary for basic functionality).
Bad Ideas:
The Drink Machine: Getting a couple of solenoids to actuate is not very challenging, and this project is quite common. Without truly exceptional industrial design and a high degree of functionality, this project usually ends up looking like keg that sort of works.
The Electrified Skateboard: There isn't anything technically challenging; you simply purchase the motor(s), ESC, and battery and assemble them. We are not impressed. A good project needs ample opportunity for things to go wrong! Try throwing in a differential or making the deck out of composites and suddenly there’s a whole lot more to talk about.
The Pancake Machine: A 2D printer with a batter dispenser is more than likely to end up a mess. If the controls and software are executed well, it could be interesting, but most often, the outcome is rather lackluster. Generally speaking, working with food can be tricky since it requires consistency of input and organic shapes aren’t precisely geometric, they’re….. organic.
A 3D Printer: You can (and probably should) buy one for $200 on Amazon. Try for something more challenging! If you’re into making tools, a CNC would be a better choice as the design space is now force constrained making the challenge substantially more difficult. When you finish most good projects, you’ll have two sizable lists: things that went wrong the first time, and things that you want to change/improve. Why not make a 3D printer/CNC combo while you’re at it!
A Quadcopter: Assuming you buy the parts off the shelf, on its own, it's not a novel project- similar to the 3D printer. However, building an EVtol or using a quadcopter as a platform for another project can be worthwhile.
Good Ideas:
The Aesthetic Chess Machine: This offers a great opportunity to create sleek mechanisms with "clever" code components that may be sourced from libraries, but the end result thinks on its own! Form factor and implementation are everything here, allowing you to execute on some beautiful and creative design choices!
A Handheld CNC Router: This project necessitates a machine that can compensate for your movements (clever code), withstand the high cutting forces, and be in a user-friendly form factor.
A Homemade Injection Molder: This project involves handling high forces, precise heating control, and producing a "finished product" that you can present and capture in photos. It also involves firsthand experience with tooling design and mold defects! Two things which any company doing plastic part design will go absolutely nuts to hear you talk about your experience with in an interview.
An Automated Fire Extinguisher: Combining a cool actuation mechanism with computer vision components for a compelling demonstration.
A Custom Carbon Fiber Bike Frame: While this project may not possess all the characteristics we mentioned, it is unique, involves intricate analysis, and demands challenging manufacturing processes, making it an exception.
Simply put, a good project is one where things can (and do) go wrong, there is a clear objective with freedom of implementation, and you learned something! A bad one is one with little to no creativity (it’s hard to design something new when the design space is so limited), is too easy (or too hard for that matter - infeasibly difficult project ideas sit on the backburner long after they should’ve been tossed), or is boring. It’s a personal project, so get personal! No matter what it is, make sure it’s something you’re interested in. If it’s not, the fight will be uphill from the beginning.
Taking Stock
Understanding your inputs is essential in constructing the framework to build a successful project. You need to choose an idea and scale that matches the resources you have at your disposal.
What tools do you have access to? Can you get access to a 3D printer, laser cutter, waterjet and/or CNC? How much volume can you handle with these tools (how many iterations could you complete for your project)? If it’s a friend working in the machine shop that’s doing you a favor, making them crank out 6 iterations of a 3 op piece would be a tad overkill. Iterate faster using 3D printing where applicable, then upgrade for the final version once the geometry and setup has been validated. This will help save money, material and favors for another day.
What is your budget? Selecting a project that requires large motors and substantial amounts of metal probably doesn't make sense if your budget is less than $200. This is typically one of the more challenging decisions as most personal projects for young engineers tend to function on shoestring budgets. We would encourage you not to pinch pennies too hard on a good project. If it’s something you’re excited about, consider it an investment. Compared to the tens of thousands of dollars a year you probably spend (or have spent) in tuition, you’ll probably learn 100x more per dollar invested on a good project (and it’ll be more valuable for your career). Purely looking at the return on investment (ROI), spending a bit more on your project if you can afford it is a deal you can’t afford to pass up. Plus, after you’re done, many of the components will be able to be reused in your next project!
If you’re especially concerned about budget, there are often grants provided by your school for projects, estate sales, or junkyards you can pillage. Repurposing parts from broken or cheap secondhand electronics can save a shocking amount of money. My personal favorite item to deconstruct and ravage for parts are used microwaves. The number of microwaves available on the street, Craigslist or Nextdoor is near infinite. In them, you can find motors, large capacitors, LEDs, a magnetron, and a high power transformer (among other things). These items can be used on any number of projects. Another popular item are the motors used in car door windows! Crack one open at a junkyard and go crazy! Oftentimes, inspiration for a good idea will strike once you have parts in hand. Getting ideas at a flea market or junkyard is a great way to do that! Plus, nothing makes a bad day good like a little deconstructive engineering.
What's the timeline? A common mistake made by many students in project classes is waiting for the professor to provide the rubric and say "go" before starting on a project. This is a significant error because the first 3-4 weeks of the semester are usually the least busy and offer a fantastic head start on a project. Don't let yourself make the excuse "Oh, I had midterms" when you're two-thirds of the way through the semester with nothing to show for it.
How many members does your team have? In reality, 1-2 team members often do the majority of the work in a team project, so having a team of 10 people is usually excessive. We'll revisit this later when we discuss how to divide the project.
CAD Software
The choice of CAD software can make or break a project. If there's one lesson we learned from team projects in college, it's that Solidworks DOES NOT WORK. You won't have PDM (Product Data Management - basic CAD filesharing software) running, and everyone will come to a meeting with CAD work done in different software (one person using Solidworks, another using Inventor, a third using Tinkercad or something random unfortunately), and after 20 minutes of downloading and uploading files to get everyone's work into the same assembly, nothing will align. The units will be off, and the parts won't fit. It's a DISASTER.
That's why we have one rule for projects: they must be done in Onshape. If you want to join our project team, you have to use Onshape. There is no other option.
Think of Onshape as the Google Docs of CAD. It's free for educational purposes and allows everyone to work in the same document. It runs in a web browser, so people can use Macs, PCs, and even Chromebooks. It's a game-changer for group projects.
The revision history works like Google Docs, allowing you to see the changes people make and when they make them. This helps with accountability. You'll know how much time people are investing in the project and if they rush to complete it at the last minute.
You can also collaborate on the same document simultaneously without overwriting each other's work. This is crucial for CAD work and collaborative efforts.
You can even leave comments and store important PDFs in the CAD workspace. As you'll see throughout the rest of this guide, all of our work was done in Onshape. The best part? It’s free for hobbyists and students! Get an educational Onshape account here.
Execution
Once you've selected a project and assembled a team, it's crucial to create a roadmap with deadlines and realistic expectations. Everyone should mark their midterm dates on the calendar to get an idea of when you'll have time to execute. With a roadmap in place, it's essential to shift your focus to risk management.
What are the unknowns? Keep in mind that there’s a big difference between not knowing how to do something, and not knowing if something can be done. The former presents only medium risk, where the latter must be addressed immediately as it could have downstream implications for the project as a whole. Keep in mind that when there are things you don’t know, there’s probably also at least that many things you don’t know you don’t know.
What aspects should you validate early to prevent going too far down an unproductive path? Are there a few different architectures you’re considering? Try doing a small sprint to evaluate the predicted merits of each. There’s not time (nor willpower) to do a full parallel path, but checking to compare a couple different architecture proof of concepts can let you quickly choose if not the best, then at least a viable path forward.
Isolate. Isolate projects into compartments with concrete deliverables. What parts of the project rely on each other? Sometimes it’s difficult to test what will or won’t work without another functional component. How do you test a touchscreen without programming the GUI? Or controls without a robot to control? These blockers can often be circumvented by creating scaled or representative models (or digital models if you’re so inclined), but keep in mind that it costs extra (potentially unnecessary) work up front, so keep project flow linear if possible.
Usually, completing a Minimum Viable Product (MVP) before midterms is crucial for a successful project. What's an MVP? It's the most basic form of your complete idea. It can be assembled with hot glue, 3D printing, and even cardboard. It doesn't need to look attractive; it just needs to demonstrate that the idea can work and provide insights that will help you refine the final product. Your MVP should serve as an integration testbed that allows for early testing of programming and electronics, rather than waiting until two days before the final deadline (which happens far too often).
One of the most significant challenges in team projects is member participation. Typically, 1 or 2 people do most of the work while others do the bare minimum to secure a grade. This is unfair but often partly the fault of the 1-2 active contributors.
Firstly, no team project should have more than 5 members. It's practically impossible for 6 or more people to work effectively on a single class project and all make meaningful contributions. For groups in the 3-5 range, the key is to divide the project into verticals that each team member can take ownership of. Each member must feel directly responsible for a deliverable. A common approach is to have 2-3 individuals focus on mechanical subsystems, one on electrical aspects, and one on software. The electrical person should make all the moving parts work independently of the mechanical system, and the software person should collaborate with this electrical testbed to mitigate code-related risks. One hotbed issue with splitting up responsibility for deliverables is trust. Making people responsible for individually critical verticals means introducing multiple points of failure. Now, someone not pulling their weight isn’t just 20% more work for everyone else, it’s catastrophic failure. I would like to say that people will rise to the occasion, and have often found that to be true, but I will also say that there is always the distinct possibility that they will not. That is why work parties and status updates are critical. Don’t let someone get away with generic statements like “looking into motors.” Make them show you the different options and talk about the benefits of each. Dive into their codebase. What functionality has been coded so far? Has it been tested? Nothing is worse than testing code for the first time with only 2 days left until the project deadline. This should be avoided at all costs.
To rectify many of the issues and risks presented above, we also highly recommend "work parties"—3-5 hour blocks where all team members are fully present and engaged with project-related work to ensure progress and status updates on all fronts. Members ideally should not be working on homework for another class, on Instagram, or studying for midterms. Each work party should start with a 3-5 minute update from each group member in which they present slides on their current work status, updates since the last meeting and any potential risks/blockers they have been dealing with. Group input may be able to resolve those blockers, and shame of presenting nothing to the group will force them to do at least some work. It’s impossible to hide in the weeds when you have to present on work you alone did, and many people tend to find this motivational to keep the ball rolling on their individual deliverables. This also mimics an industry environment in which you give technical updates in 1:1’s to your manager or to the broader team. One vital piece of being able to present on those deliverables is documentation!
Documentation
One of the most critical aspects (possibly the most crucial) of a project is proper documentation and an effective final presentation. If you didn't document it, it didn't happen. This is especially important for those who want to include the project in their design portfolio. Throughout the process, take screenshots of CAD work, photograph both successful and unsuccessful prototypes, and record your thoughts as the semester progresses. We recommend creating a mid-semester presentation document with all the updates on progress up to that point. This ensures that you don't forget about all the hard work done early on. At the end of the project, a well-structured presentation is essential.
It's all about storytelling. What challenges did you tackle? What unique design innovations did you develop? Why is the project a resounding success? Often, the grade you receive or the impression you make on an employer depends more on the presentation and framing than the actual project itself—presentation is everything!
We thought it would be helpful to break down a project we did into the steps discussed above!
The Idea
Schools have different ways of incorporating projects into their curriculum. At Berkeley, we had a Capstone class that seniors typically took. However, as juniors, we decided to enroll in it so that we would have our project ready for full-time recruiting. In the months leading up to that class, we spent countless hours brainstorming ideas and ultimately narrowed it down to three options:
HARI (Home Assistant Retrieval Itinerant)... okay, we might need to work on that name some more. We thought that building an autonomous mobility device would be exciting, and we debated whether it should be a retrieval robot or an automated walker. We never fully agreed on which direction to take.
Gift wrapping robot: You provide it with a box within a certain size range, and it analyzes the package's size to adjust its actuation and then wraps the box. The end result is a wrapped box that you can give to anyone as a demonstration of success!
A cable-actuated robot: This idea was inspired by a previous project (the automated fire extinguisher), but we wanted to scale it up and gain a better understanding of the mechanism. The challenge was that we didn't have a specific end-use idea for it yet; this would come later.
Taking Stock
Alright, so we had three team members, a budget of $600 ($200 per person), and 16 weeks to work on our project. We had ample access to 3D printing and laser cutting, as well as some access to a machine shop.
We chose the wire-actuated robot because we came up with an application: autonomous vehicle charging. We understood the importance of getting a Minimum Viable Product (MVP) working before midterms, so we started with a very modest goal. Our objective was to have a 12 inch actuated arm operational in 12 weeks.
Execution
We divided our project into three key components: Actuation, Vertebrae, and Computer Vision (CV) tracking.
Person 1 was responsible for actuation. We aimed to have two sets of moving segments, which meant we needed four servos (two per segment). Ensuring proper tension in the cables was critical for achieving fast response times. Person 1 spent a significant amount of time refining a spool mechanism with built-in tensioning. Arranging all eight strings correctly, without excessive friction, also required significant fine-tuning.
Person 2 was in charge of the vertebrae. To achieve the up-down and left-right motion we desired, it was crucial to prevent torsional movement in the “snake's spine”. Initially, we considered using numerous universal joints, but issues with set screws slipping made this approach unworkable. After conducting quite a bit of research, we switched to using speedometer cable, which is apparently manufactured specifically to have high torsional stiffness. While this solution worked well, it took multiple iterations to fine-tune the clamping and string routing.
Person 3 was responsible for the code. We wanted the snake to be able to track a colored marker (proof of concept). Person 3 invested a considerable amount of time refining filters and integrating libraries to achieve tracking functionality with a Pixy Cam. Person 3 also had to ensure the servos mapped properly to the Arduino, implementing a closed loop control algorithm. Note that nothing beyond a basic mathematical model of the snake’s travel was constructed given the deformable nature inherent in cable elements making open loop kinematics wildly inaccurate.
Integration
We didn't cover this in the guide above because this step can vary significantly depending on the project. For us, integration occurred during our weekly work parties. We would assemble the new vertebrae, connect them to the servos, and then power up the entire system to assess our progress. In the initial stages, we faced challenges with a droopy and twitching tail-like structure. However, through ongoing refinement, we eventually achieved tracking and fluid motion.
Documentation
We are huge documenters so there are literally 100+ slides showing the development of this small project, but a few highlights of our most memorable issues and checkpoints are compiled below!
Priority #1: document clearly.
Priority #2: aesthetics!
Here is our Final Slide
Then at the very end we took white backdrop photos and video. This is essential for having a nice looking portfolio! Do your best, but keep in mind it’s okay if there are shadows or non idealities in real life. The pictures can always be post-processed and any pictures are better than none at all.
Enjoy
What now?
Come up with some ideas, make an Onshape account, and get building! Hobbyists and students are the only ones who get Onshape for free outright, but businesses can get a Professional license for free for 6 months and startups can always apply for free licenses via Onshape’s startup program!