| General > General Technical Chat |
| Ideas for my Thesis topic |
| (1/3) > >> |
| flienky:
Hello everyone, I am about to finish by Bachelor's and Integrated Master's degree (5 years). This year I will have to choose my Thesis and I need your help. About me: I love programming and this is what I want to do for a living. I find embedded programming challenging and a very interesting subject compared to "regular" SoftEng. I don't see myself designing boards full time, but I understand that designing&laying pcbs will be a necessary skill. I want to find a Thesis, that will be appreciated in my first job search. The goal of my thesis is to use it (along side with other projects) to find a good first job. The Thesis that I choose would like to consist of 75% programming and 25% pcb design. Here are some ideas that I have in mind: 1) Automated testing farm with HIL testing More and more devices (iot, cars etc) are being connected to the internet and receive software updates. Firmware releases will have to be produced in short cycles and being tested automatically without any manual work. I plan to build a testing farm which will be used to test the new software updates at a more rapid pace. The farm will consist of the following: * A supervisor hardware (maybe rasberry pi) that will distribute the testing jobs to the Testing Modules * Testing Modules that consist of a (custom?) microcontroller and 2-3 IO cards. The testing module will install the software to be tested to the device (for example an ECU) and run a hardware in the loop test to check if the software update passes the tests * Α UI that developers will use to track the progress of the tests. What do you think about this thesis? Is it worth it? Maybe is it too complex? Do companies for example in the Automotive Sector (like Tesla that pushes so many software updates) use these types of test gigs to test their code? Do they implement it in-house, or buy equipment from companies like National Instruments? 2) Build a custom microcontroller. Connect multiple of them with CANBUS (or UART). One of these will be master and the rest of them are slaves. The master will communicate wirelessly with a server and receive software updates for the rest of the controllers. The master will check if the update is valid and flash the firmware though the CANBUS. This type of setup will be used widely in the automotive sector. Almost all new electric cars have OTA firmware update capabilities. Is it worth it? What do you think about this topic? ------------------- Which of the two topics would you choose or appreciate more if you were the interviewer? Feel free to comment each topic and recommend a third one! Thanks! |
| Kerlin:
Sounds like an interesting project. At this stage I wouldn't make any recommendations a on particular project, but offer one bit of advise. Make sure you can finish it. I did a really ambitious embedded project and couldn't finish it. I was also working and had a son that year, was a mature (old) age student. I took my entire one months holiday leave just to finish my project. I did complete it and pass, it was shown at the Uni's open day. Was in panic, coffee guzzling mode the whole time. Not a good way to spend a years leave. Having said that I do have a friend who couldn't finish it but wrapped it up as it was, in its unfinished state, wrote some conclusions on what he had learnt and how he would have completed it if he hadn't of run out of time. He passed. Prime directive - aim to complete it and pass. |
| MIS42N:
In my opinion [and I mean, my opinion which could be completely wrong], you are approaching this from the wrong direction. You have decided on an application and already building it in your head, and you don't know if it is useful. Would it not be better to identify a need and see if you can fill it in an innovative way. The guy who turned Microchip around met a guy who made remotes for televisions, cd players etc, and asked him what he would like most and his answer was to not have so many different products sitting on shelves. And the reason he had products was because there was a lead time on getting the specific ROMS for each type of equipment. Microchip man said if I could give you a field programmable microprocessor would that fix the problem - which it did and Microchip went from heading towards bankruptcy to a billion dollar company. Point is he identified a problem, and produced a solution. The problem came first. So if you think an automated test rig is what you want to work on, the question is (and you asked it), how is it done now? Is it satisfactory (i.e. no use reinventing the wheel) or can I (a) speed it up or (b) make it cheaper or (c) improve it in some other way. As an example (and maybe not a good one) I have been working on a GPSDO (GPS disciplined oscillator) design. My criteria was to make it use less parts than any other design (which as an outcome means it is most likely cheaper than any other). The end result is most likely the cheapest GPSDO where all of the functions normally handled by ICs are handled by software in a $1.50 processor. Yet it has more user functionality than off the shelf units costing 5 times as much. Admitted I didn't see if there was a market (I did this as a programming exercise) but now I've designed and had made a PCB there's been a stirring of interest. If it were a thesis, I'd say I'd identified a market (even though I've admitted here that I didn't), designed the hardware, used some innovative programming to address cost and functionality, fulfilled market expectations. It may not get extra brownie points when submitted, but I bet a prospective employer would like that it wasn't just the programming, it was the research that went with it. |
| thinkfat:
My advice is: don't overstretch yourself, don't underestimate the time you must spend to actually write the thesis. This may sound banal to you, but in my experience only few students understand that project management is implicitly part of the work and will be part of the evaluation. Choose your goals wisely, make a plan with realistic milestones. If you only have three months, forget about doing complex hardware projects. |
| Zeyneb:
If you can arrange with your university to do this while you are intern to a company. An internship, that would be the best option I think. Because they can inform you about what element they are currently missing in productivity, efficiency or quality. I think it is very important to spend significant effort to find a suitable company that is involved in the topics you’re interested in. When in the company you first need to see how they are doing things right now before you can formulate what the exact topic is going to be. And while you are going they can guide you regarding what constraints your design needs to have to make it suitable in their business process. |
| Navigation |
| Message Index |
| Next page |