Developing a Research Question

Introduction

Robotics as a field is highly multi-disciplinary. Even within that field, designing a robot requires knowledge and input from a variety of domains, including biology, materials, mechanics, mechanical design,dynamics, CAD, manufacturing, optimization, control, etc. Because of its highly interdisciplinary nature, as new knowledge emerges from any of its connected fields, there is an opportunity to innovate. Designing novel and innovative robots requires adapting new ideas from those neighboring fields, connecting them together, and remixing them with existing techniques, in order to realize new modes of locomotion or achieve higher performance.

This is where research comes in. The goal of research is to help us understand the underlying mechanisms that connect these concepts together. Unlike industry, that may be more focused on applying the output of knowledge to solve an applied problem, research often helps us make connections and understand relationships between concepts

Research Questions

Research questions are at the heart of this process. There can be big research questions or small research questions. In fact, the biggest research questions, the answers to life, the universe, and everything1 are composed of a thousand small questions.

In fact, this class – and my research in general – focuses on a small set of research questions. Research questions are at the heart of every paper and proposal I write. Asking a good research question is important because as engineers and scientists, it is important to understand underlying connections between two disparate domains before trying to connect them together for an application.

The biggest questions – the answers to life, the secrets of intelligence, and our understanding of matter – These by nature are open ended, multifaceted and, in my opinion, have don’t have a single answer. But these are terrible questions to start with when opening up a new line of questioning. Why?

On the other hand, questions like these can be broken down into smaller problems. Take what I might consider the driving question of much of my research:

How can higher-acheiving robots be made more easily and more affordably by more people?

This remaines a poor starting point for new projects because it still fails by the same points above, except that maybe my expertise can go a long way towards answering that question, in the longer term. Given my interests, how can I break that question up into smaller parts?

A good research question has the following characteristics:

Achievable / Tractable

The question should focus your work towards an achievable goal. It should be limited in nature and may even hint at the approaches you might use to answer it. Consider that for this class, an appropriate research question must be able to be answered in one semester.

If, after some background reading and initial testing you find it’s too broad, you should adjust and refocus your question. This can be done using a number of techniques:

Novel

Through background reading of prior research, existing patents, and popular literature you should be relatively sure that the question hasn’t already been answered. Granted, if it’s a good idea, other people may already be working on the problem, but if the main structure of the answer is already mapped out, then, unless you are reasonably sure you have something new to add, you will probably find that most of the most interesting questions within that subfield have already been asked and answered as well.

Though it is hard to prove a negative, a five minute search on google scholar by a domain expert like your professor shouldn’t come up with many highly cited results. The toughest part is identifying the keywords that people within the field (if it already exists in a nascent form) already use.

Here are some ways to make a research question more novel

Interesting, Timely, and Relevant

Your question should be interesting to the broader community. How do you achieve this sweet spot of something that is interesting while at the same time being novel? This is the challenge to research in general – finding topics that are unexplored while at the same time of interest to the community. If it is truly of interest, probably someone else is already thinking of your idea. If someone else has solved it already, less can be learned from you re-examining the same question. The only way to work through this is to proceed quickly and efficiently through your research, and to publish your results early and often.

Open-Ended

Most research questions should contain “how” or “why” phrases, and probe more fundamental relationships. Why? Yes/No questions are boring, and most of the time are not fundable. Open ended questions are the bedrock of science, while application-focused projects are typically more goal-oriented. Goal oriented questions are great in a number of scenarios, especially in military and industry-oriented projects, but to get funded by the National Science Foundation, your project needs “Intellectual Merit”, which relates back to answering fundamental scientific questions of interest to the community. So let’s practice

Exercise: Turn this goal-oriented question to a more open-ended question

Use some of the techniques described above to refocus the given research questions to something more open-ended

Goal-OrientedOpen-Ended
Can I make a jumping robot with liquid crystal actuators?What are the optimal material design parameters to increase the performance of jumping robots
How do I make the world’s fastest robot?2

Modular

While the research question you ask should be rational, achievable, and focused in order to fit within the constraints of a single project or semester, your question should also be able to be assembled into a broader picture. This helps you as a researcher create a unique narrative around which you can define your academic career. This is useful for a number of reasons. First, building up blocks of questions defines you as a domain expert. People start to associate you with a sub-field, or field, and – if it is interesting to the community – you will probably become more sought out to speak, collaborate on proposals or papers, etc.. Second, it gives you a mastery of a research pipeline that allows you to ask newer, related questions better and faster. Like the gravity in a black hole, the better you are at something, the better you become, and it accelerates. Finally, having multiple pieces that you can fit together in a variety of ways allows you to reframe your expertise as the field evolves. While some topics become “solved”, if you can swap out one piece of your research for another, you can continue to remain relevant, even if some of your topics remain stale. Who knows, maybe it will become popular again, if something new comes along and makes it relevant again?

In summary the process of asking new questions can help you build up a research portfolio that you can remix and reframe as needed.

Leverages your own abilities.

This may be obvious, but you should ask research questions that are both interesting, and that you are best able to answer. If you are a mechanical engineer asking a research question about game theory, you may fall into some of the above pitfalls of novelty, interest, and focus

Demonstrates your expertise

Finally, a good research question also demonstrates you know enough about your field to make a novel, interesting, and relevant new contribution to the field.

Relevant

And finally, the question you ask must be answerable with the topics you learn in Foldable Robotics.


  1. Douglas Adams ↩︎

  2. This is s`till goal oriented. I said how, but I don’t care why. ↩︎