🎙️ Podcast Link 🎙️

Last week at the 2024 IEEE/RSJ International Conference on Intelligent Robots and Systems (#IROS2024) in Abu Dhabi, Cyrill Stachniss and I were invited to give a tutorial on “Writing Papers People Love to Read” by Julie Stephany Berrio Perez on behalf of the IEEE Young Professionals.

Now, to give a proper, comprehensive treatment of all the ins and outs of paper writing really requires something more like a full day workshop, and it’s something that researchers in our teams get exposed to gradually over a period of years.

On this occasion, Cyrill and I had around 20-25 minutes each, so we focused on some of the key concepts, techniques, tips, tricks and pitfalls, followed by some great discussion and Q&A with a great turnout for the final formal event on day 2 of the conference.

Cyrill went first, providing a bunch of tips and suggestions on paper structures, what should go into each part of the paper, nailing your contributions, grabbing and keeping the reader engaged and general todos and not to dos. You can check out Cyrill’s talk here: https://lnkd.in/dywR5rjp

I then followed up with a tutorial on how to think about the art of paper writing, with a number of high level concepts – including the “must dos” and the “field and audience-specific factors”, explicit ways to think about and categorize your paper in order to enhance your writing, presentation and a paper writing superpower – multi-step anticipation – where you repeatedly anticipate and pre-emptively respond to the likely questions or thoughts what will occur to your reader (again, best done with an experienced writer and researcher).

Common to both of our talks was the incredible value of “pair writing” early in your career with an experienced writer, where you sit side by side and write (type) together – the learning rate is phenomenal when you get this right, far far better than at-arms-length feedback via “marked up draft” or general comments in an e-mail or slack channel.

Common to both of our talks was the incredible value of “pair writing” early in your career with an experienced writer, where you sit side by side and write (type) together – the learning rate is phenomenal when you get this right, far far better than at-arms-length feedback via “marked up draft” or general comments in an e-mail or slack channel. In Cyrill’s case, he worked with Wolfram Burgard, and in my case, I worked with Gordon Wyeth.

As well as the general theme of respecting your reader’s precious and sometimes not-entirely-focused time and energy, and all the things you can do to maximize their enjoyment and clarity whilst reading your paper, and minimize their confusion and aggravation.

Again, these are high level tastings of what are a bunch of concepts that can go much deeper and into more detail.

Thanks to Julie Stephany Berrio Perez for very nicely filming and editing this together.

#paperwriting #robotics #research #conference #academic #publishing #papers #communication #writing #comms #university #PhD #postdoc #academia #IROS

Full Video Notes

Today, I want to go over five or six key points—some high-level concepts about writing academic papers. Over about the same period as Cyrill has been writing papers, I’ve been doing the same, and I want to share some thoughts on approaching paper writing as a science. We’ll cover the different types of papers and their outcomes, a concept I call the paper contribution pillars, writing and paper development, presentation (reiterating some of the excellent points Seral made), and finally, I’ll close with something I regard as a “superpower” for writing. This superpower—multistep anticipation—applies not just to papers, but also to job applications, promotion applications, grant and fellowship proposals, and basically anything where you’re proposing ideas or initiatives. I’ll explain it toward the end of this presentation. Much of this material is available on my website and YouTube channel, Hacking Academia, so feel free to check it out if you want more detail on some of these concepts.

Let’s begin with a quote. I thought this was by Abraham Lincoln, but apparently, it’s by John Lydgate: “You can please some of the people all of the time, you can please all of the people some of the time, but you can’t please all of the people all of the time.” This applies directly to writing academic papers. No matter how brilliant your paper is—whether it’s transformative or award-winning—there will always be reviewers and readers who are dissatisfied. So, don’t sweat it too much; perfection is impossible. However, even with that caveat, there are universal principles and context-specific strategies you can use to make your paper compelling, enjoyable, and effective.

There are a set of almost universal principles that apply to paper writing, regardless of the discipline, audience, or your personal style. These are the no-brainer “always do this” strategies that Seral covered so well, and I won’t repeat them here.

Beyond universal principles, there’s a second tier of strategies that depend on context—the audience, discipline norms, and even your own personality. Robotics is a broad discipline with varying norms across its subfields, and what works for one type of paper might not work for another. While a formulaic approach can get you started, it’s essential to tailor your paper-writing process to your unique circumstances.

For example, knowing your audience is critical. If you’re publishing across disciplines—say, robotics, machine learning, or computer vision—you’ll notice that the norms for writing papers differ significantly. A paper for a top computer vision conference will look very different from one submitted to Science Robotics or Robotics and Automation Magazine. Understanding and respecting these norms is a key part of crafting something that resonates with your target audience.

Another important concept: you need to master the basics before moving on to advanced techniques. Writing a great paper isn’t about flashy methods; it’s about building on solid fundamentals. Think of it as a triangle: the basics form the base, intermediate skills build on that, and advanced techniques come at the top. If your fundamentals aren’t strong, the most sophisticated strategies won’t matter.

Readers approach papers in various states of mind and settings. Some might be reading in a quiet, focused environment, while others are skimming your work on a plane or in between meetings. Reviewers are often tired, rushed, or juggling multiple responsibilities. That’s why it’s critical to grab their attention early. Your paper should make its key points immediately clear, even to someone who’s not devoting their full cognitive energy to reading it.

Great papers distill a vast amount of background research, experimentation, and ideas into a concise, focused document. It’s like the “iceberg principle” of novel writing: most of the work—background experiments, insights, and context—remains hidden beneath the surface. Only the sharpest 1-2% makes it into the final paper.

As you’re preparing your paper, think consciously about its type. For instance, is it a “better-than-before” paper that demonstrates incremental improvement on a popular technique, like SLAM? Or is it a completely novel contribution, where benchmarks don’t exist yet? Your paper’s type will influence how you communicate its contribution.

You can also think about your paper in terms of contribution pillars: Theory – A fundamentally cool or innovative idea. Experimental Results – Stunning empirical evidence. Effort Contribution – A massive dataset, simulation, or training environment. Most papers are strong in one or two of these areas, but rarely all three. Be clear about where your strengths lie and focus on those.

Specificity is crucial in academic writing. For example, rather than claiming your method “outperforms previous work,” specify how it’s better and in what context. Readers are less likely to react negatively when you communicate the key specific benefits of your work. Similarly, being objective about your paper’s weaknesses—whether theoretical or experimental—builds trust with your audience. It helps readers understand how to build on your work and ensures that your claims are both credible and constructive.

The writing process often involves collaboration. Sitting down with an experienced writer—like your supervisor—and co-writing a paper can be incredibly formative. While it may feel disheartening to have your words heavily edited at first, it’s a fast-track to improving your writing.

A useful strategy is to draft your abstract and contributions section before starting the actual work. As your research progresses, revisit and refine these sections. This acts as a reference point, helping to shape both your thinking and the eventual paper.

Presentation is critical. Reviewers should never have to struggle with formatting, grammar, or clarity. These are entirely within your control and represent easy wins. Similarly, invest time in creating compelling visuals. A perfect front-page figure—one that succinctly captures the problem, methodology, and outcomes—can elevate your paper and become a reusable asset for years to come.

Finally, I want to talk about what I consider a “superpower” in academic writing: multistep anticipation. This involves predicting what readers or reviewers will think as they progress through your paper and preemptively addressing their concerns. It’s about going beyond a single step of rebuttal to anticipate and respond to multiple layers of feedback.

For example, if your paper is on SLAM, a reader might think: “SLAM is solved; why is this important?” Anticipate this by explaining a specific use case where SLAM remains unsolved. Their next thought might be: “Okay, but why not use a pre-mapping or localization approach?” Address that by showing why those alternatives don’t work in your context. By preemptively tackling these objections, you strengthen your paper and reduce the need for a rebuttal later.

Writing papers is an iterative process, and while it can be challenging, it’s also deeply rewarding. By mastering the basics, tailoring your approach to your audience, and applying strategies like multistep anticipation, you can craft compelling, impactful work that resonates with readers. Thank you to the workshop organizers, and feel free to explore more resources on Hacking Academia. I look forward to chatting further—thank you!