Different students enter the PhD program with different backgrounds. Some students take research-oriented modules (courses in US) at undergraduate level. Some other do a research masters before doing a PhD. However, the kind of research questions we address in a PhD are very new and requires a long period of deeper investigation. Therefore, it is important to know how to find a good question that gets you excited.

Direct encounter: Usually, a good question comes from an experience. In my case, I experienced how hard it is to derive the dynamics of a robot with high degrees of freedom (DoF). I actually tried to manually derive dynamics of a 4-DoF manipulator called Mark-II from Yasakawa Corporation, and then ran a Mathematica program to do a symbolic derivation for a 7-DoF robot manipulator called PA-10. I experienced how long the equations grew and thought how the brain might be dealing with a body of about 37 DoFs for model based predictive control. This direct encounter with the problem is very important, because it gives you a cause to work towards.

Look around: After finding a problem worthy of addressing, look around to see how others have approached to solve it. This is where you will see different schools of thought. Be careful. There are glaring band-waggons out there. It is so tempting to get in one of them. Don’t blindly follow them unless you have a good reason. Usually following is tiring. Think carefully trying out simple derivations and doing simulations or even doing simple physical experiments to see what kind of approaches get you excited. Some approaches appear very exciting, but direct usage will prove to be not so effective. At this point, it is very important to consult your supervisor. The supervisor may have a favorite approach. Most experienced supervisors are open for change and a good reasoned discussion will help you to benefit from their experience to polish up your research question and the method you want to address it. You should always check if there are quantifiable methods to address your research question. For instance, if you want to test whether there is a particular class of mechanisms available to minimise the size of collision force when a robot is dropped from a height, you should think about testing methods, candidate mechanisms, and the range of design paramaters to assess the scope of analysis. Sometimes, your laboratory may not have the full capacity to help you. This is where you can look for collaborations. Try to reach this level of planning logistics within the first 4-6 months in your PhD.

First experiment is important: Once you know your cause for the PhD and once the approach and collaborations are established, break your approach down to smaller steps. Don’t worry too much about how the last experiment will be done. Worry about your first experiment. Distill out a refined research question that needs a novel answer that you can reach in about 6 months. This is important to boost confidence. Temptations will be high to find the ultimate answer to bring your field to a conclusion, but even in that case, it is important to make a first firm step. In this first step, master the tools and techniques involved in your field. In my lab, students take this time to master robot design and fabrication skills, coding skills, data analysis skills, and cool math you can use to solve difficult problems. Develop the habit of reading at least one paper a week that empowers you with powerful tools to solve problems.

Documentation: It is important to develop the habit of keeping things in a well sorted file structure. Open a folder for each project. Have sub-folders for data, reports, codes, papers you read (using a repository like Madeley is also great), designs, and other resources. This is going to save time when you write a paper at some point. Now you have cloud resources like Box and Onedrive. Back up everything securely.

Writing the first paper: If everything works out, after about one year into the PhD, you will have some new results worthy of publishing. Sometimes, the first attempt doesn’t work out. But all failed attempts teach us lessons. Don’t get discouraged if the first experiment doesn’t work out. Develop the resilience to come back with a different approach or to formulate the question in a different way. Then when you write the first paper, you will have comparative results. The importance of reading papers at least one per week is that in 6 months, you would have read at least 25 papers. This is enough to write your first paper. Start writing why the question you addressed is new and important, and back it up with papers you read. Write down your methods very clearly keeping in mind that somebody should be able to read your paper and be able to replicate it for independent verification. Results and interpretations need to be as sharp and consistent as possible. Plan to go through several rounds of revisions with your supervisor and lab mates before any submission deadlines. I ask my PhD students to have the paper in a reasonable level for revision at least one month before the deadline. Have this as a ballpark period for revision in your first paper. This is the time where you develop the skills of articulating a concept clearly, present it to an audience, receive criticisms, and develop good habits of critical reflection.

Completing the cycle: You will of course get review feedback. Some suggestions I have given in this note can be useful to go the rest of the journey. Once you get your first paper published, you will have your next research questions coming up easily. The advantage of taking an approach you are passionate about to serve the cause you selected is that it will naturally line up the next set of questions and methods you should be pursuing. My advise is to go through this full cycle of raising a question to publishing results at least 3 times during your PhD. It will give you a seasoned experience of the art of formulating good research questions.