I actually finished off my mech eng thesis last year.
You find a topic you're interested in doing and contact the lecturer in charge of it to be your supervisor. Sometimes you can propose your own topic to a lecturer in the relevant field of research.
Sometimes they want you to have a high wam as the topic can be challenging like researching how to mine an asteroid.
Once you sorted out your supervisor, just rock to him and he'll point you in right direction what to do. Normally you spend the first half reading a bunch of papers relevant to your topic, doing lit review, planning experiments and maybe sorting out materials for experiments and play around with the equipment. Each supervisor is different, heard some are almost impossible to reach, some just let you do your own thing if you don't ask for guidance. Mine was pretty easy to schedule a meeting with weekly, point me in right direction after every step and what to do when my experiment stuff up.
During the next sem, you try to start experiments but realise there's a hurdle of red tape you gotta go through and stuff just doesn't goes as plan with other class using equipment, things going missing. You try to write up all the stuff you been doing and analyse your results. Most people including me regret not doing more work in the first semester of it. You do some experiments and get results that you struggle to explain or convert into explainable data. I was in the lab 3 full days a week at the last 3 weeks of my thesis, most of it was waiting around for the testing equipment. After you spent an entire week formatting and putting your 70-100 page thesis together to submit and hope you pass. You gotta do a presentation of your thesis in front of a few lecturers and 10 or so other thesis students who are also marking you.
This was what happen with a research thesis at least. I think you can do industrial thesis, simulation thesis or even practical thesis.