David Filliat

About

David Filliat is a prominent robotics researcher whose work spans visual simultaneous localization and mapping (SLAM), continual learning, and autonomous robot navigation. Based at ENSTA Paris, he has made foundational contributions to how robots perceive, map, and navigate complex environments using vision-based techniques. Filliat is perhaps best known for pioneering bag-of-visual-words approaches to place recognition and loop-closure detection, with his 2008 paper on fast, incremental loop-closure detection accumulating over 515 citations and becoming a landmark reference in the SLAM community. His earlier work establishing frameworks for map-based mobile robot navigation (2003, cited over 220–236 times across related publications) helped define the conceptual landscape of the field. More recently, Filliat has emerged as a leading voice in continual learning for robotics — addressing how systems can learn sequentially without catastrophically forgetting prior knowledge — with his 2019 survey attracting nearly 500 citations. His 2018 overview of state representation learning for control (295 citations) further demonstrates his ability to synthesize emerging research directions with broad impact. With work spanning 3D semantic scene understanding, visual saliency, and topological SLAM, Filliat's research consistently bridges theoretical machine learning advances with practical robotic applications, making him an essential reference for students and researchers in intelligent autonomous systems.

Research Focus

Key Achievements

Top Papers

1. 1
   Fast and Incremental Method for Loop-Closure Detection Using Bags of Visual Words

   515 citations · 2008
2. 2
   Continual Learning for Robotics: Definition, Framework, Learning\n Strategies, Opportunities and Challenges

   492 citations · 2019
3. 3
   State representation learning for control: An overview

   295 citations · 2018
4. 4
   Map-based navigation in mobile robots:

   236 citations · 2003
5. 5
   A visual bag of words method for interactive qualitative localization and mapping

   235 citations · 2007
6. 6
   Map-based navigation in mobile robots:

   220 citations · 2003
7. 7
   Real-time visual loop-closure detection

   105 citations · 2008
8. 8
   Incremental vision-based topological SLAM

   84 citations · 2008
9. 9
   SnapNet-R: Consistent 3D Multi-view Semantic Labeling for Robotics

   82 citations · 2017
10. 10
    Environment exploration for object-based visual saliency learning

    78 citations · 2016

Key Collaborators

Contact & Links