Skip to content

Part 1: Research Prerequisites

Outline

  • How to do research

  • BLUF: Implement SOA methods + Methodically Change Them

  • Given a specific problem

  • Learn about how others have approached the problem
  • Implement a state of the art approach
  • Characterize / benchmark the problem in the SOA method
  • Propose / implement candidate strategies for fixing the problem
  • Test strategies and iterate until meaningful improvement is made
  • Write findings outlining the above steps in conference or journal

  • Publish --- submit + defend against reviewers

  • How to generate good research ideas

  • BLUF: Understand the problem space

  • Step 0: ChatGPT

    • Prompt: "What are the open state-of-the-art research problems in spacecraft guidance, navigation, and control?"
    • Prompt: "What are the open state-of-the-art research problems in astrodynamics and celestial mechanics?"
    • Prompt: "What are the open state-of-the-art research problems in scientific machine learning?"
    • Prompt: "What are the open state-of-the-art research problems in deep reinforcement learning learning?"

  • Step 1: Read popular science + news articles

    • What are the major discussion points on space.com
    • Debris
    • Constellations
    • Earth Imaging
    • GNSS
    • Read
  • Step 2: Look at sessions at recent conferences to identify research themes
    • Read multiple papers from the session
    • Iteratively refine your knowledge when you encounter new concepts / words
    • Compile a list of cited problems from their introduction / background
  • Step 3: Implement a state of the art approach
    • Be able to explain the ins and outs of the entire stack
    • Ensure that the results that you are getting match that of others
    • Use results as baselines from which you are trying to improve
    • Identify parts of the methods that are either:
    • a) suboptimal (slow, made assumptions)
    • b) don't exist (improvements that could be added)

  • Step 4: Down-select based on highest impact modifications

    • Build idea into a family of experiments
    • Connect candidate improvement to the high-level idea / problem space
    • Synthesize and you have a research problem

  • How to Manage a Research Project

  • BLUF: Generate a suite of questions that you want answered + intermingle learning with experimentation

Overview

Research is as much of an art as it is a science.

You will develop your own style over time, but there is generally accepted techniques and processes that will increase your productivity that we outline below.

Research Notes

During your degree, you will be working on open-ended problems that require many weeks or months to solve. It is essential that you keep clear notes while you chip away at your problem.

The most common way to do this is to keep a research notebook. This can be physical or digital, but it should serve as a daily, living document for which you can distill your thoughts and problem-solving.

I generally recommend the following process for your research notebook:

Research Notebook Format

- Start a new note every day
- Review / recap where you were previously (in written form)
- Define your goals for the day --- making as many quantifiable as possible
- Propose a rough timeline for completing the activities
- Make a heading for each task, and take detailed notes regarding your progress
- At the end of the day, write a short summary of what you accomplished

Individual Meetings / Reporting

In addition to taking daily notes in your research notebook, we will also meet regularly to discuss progress, set milestones, and troubleshoot existing problems in individual advising meetings. Every manager has their own preference for how these meetings are run. Below I outline my expectations.

Format of Advising Meeting

- Recap of past meeting
- Status on said progress
- Current obstacles 
- Progress on said obstacles 
- Requests for feedback / help
- Establish goals + deliverables for next meeting

Communication

Contrary to common thought, earning a PhD is not only about advancing the state of the art, it is also a statement about one's ability to communicate.

Earning a PhD requires two deliverables: a dissertation and a defense. Both of these are evaluated based on the candidate's ability to communicate their technical contributions in written and oral form. Therefore a key skill that must be matured during your time in the program is your ability to communicate clearly.

As such, when you are writing your research notes, or preparing for an advising meeting, be sure to practice good communication skills. Good communication is not a natural skill, it's one that requires constant refinement and practice. Think of it like an instrument. A little mindful practice everyday will dramatically improve your research throughput.

Proposed Communication Practices

- Write down your thoughts in bullet form
- Distill into complete sentences
- Read out loud, paying attention to the cadence (pause after each sentence)
- Have someone else read your prose and highlight confusing passages
- Have a text-to-speech model read your work back to you