In the intricate dance of research, deciding where to start can feel like standing at a crossroads. Do you dive straight into the nuts and bolts, or do you begin with a bird's eye view, sketching out the grand vision before getting your hands dirty? This is the quintessential debate between the bottom-up approach versus the top-down approach in research methodology.
Let's break it down, starting with the bottom-up approach. Imagine you're building a puzzle; rather than starting with the picture on the box (the end goal), you begin with the individual pieces. This is what a bottom-up approach feels like. You delve into the technicalities, the nitty-gritty of tools, technologies, and methods. It's like starting with the hows before you fully understand the whys. By piecing together these components, you gradually form a bigger picture or theory. It's a hands-on, detail-oriented approach that can lead to unexpected discoveries.
On the flip side, we've got the top-down approach. This is where you start with the picture on the puzzle box. You have an overarching idea, strategy, or concept—and your job is to figure out how to bring this vision to life through technical means. It’s about seeing the forest before the trees. You outline the broad strokes of your research first, identifying the symbolic concepts or paradigms guiding your journey, and then systematically break them down into technical tasks and skills.
Both paths offer unique advantages. The bottom-up approach allows for flexibility and innovative detours, as you're not strictly bound by a predetermined theory or concept. It's a playground for the technically minded, where hands-on experimentation can guide theory development in unexpected ways. Meanwhile, the top-down approach provides a clear direction and framework from the start, which can be particularly beneficial in complex or multidisciplinary research areas where maintaining focus and coherence across different subfields is essential.
Choosing between these approaches isn't about right or wrong; it's about aligning with your project's nature, your personal style, and the goals you aim to achieve. Some researchers blend the two, alternating between bottom-up experiments to inform their theory and top-down planning to refine their practice. The beauty of research lies in its diversity of methods, and understanding these approaches is key to carving out your path in the complex world of inquiry.
I am in China from May 11 to 26 to give classes at Nanjing Tech University as part of the double degree in vision and robotics that we have developed together. The students spend a first year in China, where 1/3 of the classes are given by UB teachers, then move to France for the two years of the VIBOT master's degree.
Nous avons initié plein de belles choses ces dernières semaines :
soumis un projet PHC Hibiscus avec nos collègues de l'UTP en Malaisie ;
soumis un projet H2020 avec nos collègues luxembourgeois, espagnols, italiens et bulgares ;
déposé une demande d'accréditation à la CTI ;
reçu un retour positif de l'ANRT pour une thèse CIFRE.
En mai, je pars à Nanjing (Chine) pour 2 semaines de cours de vision par ordinateur et de robotique.
En juin, nous accueillons la CTI, organisons les VIBOT Days, partons en Italie pour les MaIA Days...
Premier conseil scientifique du laboratoire ImViA en juillet.
I will give a research seminar at the SnT (Université du Luxembourg) on Friday 11 at 11AM. It will be entitled: “Vision and robotics for agriculture: state of the art and challenges”.
Abstract:
Agricultural robotics brings new challenges to computer vision and robotics researchers. First, because it requires the robot to operate outdoors on sometimes difficult terrain and in non-cooperative conditions: mud, rain, fog, sun, etc. Moreover, if it is possible to build a factory around industrial robots, it is necessary to adapt agricultural robots to the very specific context it serves, to practices and know-how. Finally, environmental and health issues will become increasingly important in the agriculture of the future and robotics will have to meet them. Our work, with a particular focus on viticultural robotics, will address some of these issues and propose robotic solutions to:
– have a better knowledge of the plant's development and health status;
– adapt the treatments applied to the plant to this state by seeking to reduce its volume (more for ecological and health reasons than for economic reasons) – in a so-called sustainable agriculture approach;
– have an objective measure of the effectiveness of these treatments by monitoring the state of the plant.
Our project for a new research centre has been approved by the French Ministry of Research. From 1 January 2019, the ImViA laboratory (Imaging and Artificial Vision) will be created under the reference EA 7535.
It will be composed of three teams:
CORES (COmputer vision for REal time Systems) in Dijon
IFTIM (Functional Imaging and Medical Image Processing) in Dijon
VIBOT (VIsion for roBOTics) ERL CNRS 6000 in Le Creusot
The story keeps going....
Busy yet interesting week. I will be in Evry for the “HDR” defense of Jean-Yves Didier tomorrow and in Dijon for the PhD defense of Richard Macwan next Friday.
Oh! By the way, the ministry recently approved our new research lab proposal: on January, ImViA (for “Imagerie et Vision Artificielle”) will be officially created under the reference EA 7535.
I will fly to Toulouse today for the PhD defense of Philippe-Antoine Gohard that will be held at the LAAS tomorrow. Back to Le Creusot on Monday!
L'équipe VIBOT animera un stand au Village des Sciences au Creusot ce weekend. Si la vision par ordinateur et ses applications à la robotique vous intéresse, c'est là que ça se passe...
I am moving the VIBOTians community (group of VIBOT/MaIA students and alumni, BSc and MSc) from Google+ (which will be closing soon) to LinkedIn. Seek the group on LinkedIn and submit a subscription request if you are a Vibotian.
I'm in Madrid for #IROS2018 all this week. You can follow our Instagram @vibot_hq for photos and videos of the event.