Thread by @maththrills

Practical tips for writing #robotics conference #papers for venues like #ICRA & #IROS that get ACCEPTED: a 🧵.

🙏Please retweet if useful!

Video:

#QUT @QUTRobotics @sourav_garg_ @TobiasRobotics @Stephen_Hausler
@ieee_ras_icra #ICRA2023 #AcademicTwitter

1) Know what type of paper you're writing: a) a robot doing better at some well known task or b) a completely new and novel task / application / area.
For a) Show your delta over previous contributions, theoretically and/or practically
For b) Sell the relevance to robotics

2) Know your context: demonstrate a detailed and nuanced understanding of the application context in which your research is situated: the more specific and up-to-date, the better

3) What are your key contributions? Be as specific as possible, broad statements are not helpful. Specificity can be around type of performance improvement, theoretical contribution, new sensoring modalities: be specific.

4) Research gap: positively and constructively identify the research gap you're addressing, without unnecessary punching down on previous research. Previous research has made good contributions, but there are still challenges to address - that's what this paper is about

5) Don't die on any unnecessary hills: make the claims that are critical to justifying the research in the paper, and nothing more. It's not a general soapbox: you are motivating the specific and usually narrow research direction you've taken in this specific paper.

6) Know the 3 main types of contributions: 1) amazing new theory, 2) breakthroughs in performance capability / results, or 3) a self-evidently awesome new capability. Strength in one can compensate for weakness in another. But mediocrity across the board is a hard sell...

7) Experimental setup: real robot if possible, dataset from real robots otherwise. Experiments should convince reader that your result wasn't a fluke - typically done with multiple experiments or datasets that vary substantially. One flagship experiment, others can be smaller

8) Candidly show your weaknesses: help researchers who follow your work understand where the remaining gaps are, and make it easy for your reader to understand not just the strengths of your approach but its shortcomings. Usually much appreciated by reviewers.

9) Near submission: the paragraph by paragraph test. Go through para by para, check each para has a single clear and coherent message, that is critically needed for the paper. Reword / break-up / remove as necessary. You should be able to summarise each para in a short sentence!

10) Have fun! Paper writing is an integral part of the research process. It isn't just an end output but is also formative in helping you think about and position your research, and identify what you could do next. Good luck!