Pierre-Louis Guhur

While you can ask a 5 year old child to bring you a spoon, it remains challenging for a robot. My work focuses on how robots can learn to navigate, to manipulate objects, and to apprehend their environment using natural language instructions. My work outperforms state-of-the-art results over 10 challenges. 6 research papers are published at ICCV21, CVPR22 (oral), NIPS21, ECCV22, NIPS22 (submitted) and CoRL22 (submitted). I am also a reviewer at CVPR and ECCV. My PhD is supervised by Cordelia Schmid and Ivan Laptev. I am a part of Inria and ENS. Aside, I am in charge of the team computing cluster at Inria Paris, and I am in charge of three classes in Reinforcement Learning, Data Science for Business and Web Development (React).

Better Vote (Mieux Voter) wants to improve our democracies by asking a simple question: are we sure to elect the best candidate with our ballot system? Actually, several centuries of research on social choice and game theory has demonstrated plenty of limitations of the widely used ballot system: voting strategies, abstension, mistrust, among others I facilitate the integration of new ballot systems in political parties (e.g. En Marche), cities (e.g. City of Paris), NGOs (e.g. UNESCO) or events gathering several 100k people (Primaire Populaire, VoterPourLeClimat.fr, mieuxvoter2022.fr, LaPrimaire.org). I am been interviewed by the French National Assembly and several radio channels. I am acting as a technical director with the help of our growing community. For instance, our open source voting platform has been used by more than 100k people (>10k votes).

MV ROBOT produces robots to clean surfaces. We developed a patented solution for autonomous vehicles based on state-of the-art methods. Robotics is the best solution to guarantee the cleanliness of your environment.