Avoiding the Engineering Trap: Focusing on Scientific Contributions in Robotics
Last week, a PhD student received a review for his paper stating, "In its current form, it seems more like a technical report rather than a scientific publication." I couldn't help but agree. In robotics, it is challenging not to stray from scientific research into engineering. When conducting research in robotics or developing a novel scientific method, it's often necessary to demonstrate that it works in the real world, which usually requires collecting real-world data. To achieve this, you need a fully functioning setup, which can take an incredibly long time, especially when dealing with multimodal data such as vision, language, gestures, robot motion, and so on. After investing so much time in getting everything to work, PhD students (and not only them) are often eager to present the extensive effort required to set up their experiments. However, they sometimes forget that this is where their research should truly begin, and where they should start showcasing the...
