Benjamin Ramtoula

Research

I am currently working on measuring domain gaps between datasets of images. I want to develop new general measurement methods and study their applications to problems in computer vision (e.g., having better metrics for generative models) and robotics (e.g., selecting maps to localise in). Previously, I also worked on multi-robot systems, SLAM, and object pose estimation.

Fantastic Features and Where to Find Them: A Probing Method to combine Features from Multiple Foundation Models Benjamin Ramtoula, Pierre-Yves Lajoie, Paul Newman, Daniele De Martini NeurIPS, 2025 We design an adapter that can probe representations from different layers of multiple foundation models.
3D Foundation Model-Based Loop Closing for Decentralized Collaborative SLAM Pierre-Yves Lajoie, Benjamin Ramtoula, Daniele De Martini, Giovanni Beltrame RA-L, 2025 We integrate 3D foundation models to improve loop closure detection in collaborative SLAM.
VDNA-PR: Using general dataset representations for robust sequential visual place recognition Benjamin Ramtoula, Daniele De Martini^, Matthew Gadd^, Paul Newman ICRA, 2024 We make a small adapter to use Visual DNAs for robust sequential visual place recognition.
What you see is what you get: Experience ranking with deep neural dataset-to-dataset similarity for topological localisation Matthew Gadd, Benjamin Ramtoula, Daniele De Martini, Paul Newman ISER, 2023 We experiment with selecting the best maps for localisation by applying domain gap measurement techniques between deployment and mapping images.
Visual DNA: Representing and Comparing Images using Distributions of Neuron Activations Benjamin Ramtoula, Matthew Gadd, Paul Newman, Daniele De Martini CVPR, 2023 project page paper code bibtex We propose a way to represent images so that datasets can be compared with customised distances.
Towards Collaborative Simultaneous Localization and Mapping: a Survey of the Current Research Landscape Pierre-Yves Lajoie, Benjamin Ramtoula, Fang Wu, Giovanni Beltrame Field Robotics, 2022 A survey of the state of collaborative SLAM.
NeBula: Quest for Robotic Autonomy in Challenging Environments; TEAM CoSTAR at the DARPA Subterranean Challenge Ali Agha, Kyohei Otsu, Benjamin Morrell, [...], Benjamin Ramtoula, [...], Luca Carlone, Joel Burdick Field Robotics, 2022 Presentation of our team's solution for the DARPA SubT Challenge.
On the Communication Requirements of Decentralized Connectivity Control: A Field Experiment Jacopo Panerati, Benjamin Ramtoula, David St-Onge, Yanjun Cao, Marcel Kaufmann, Aidan Cowley, Lorenzo Sabattini, Giovanni Beltrame DARS, 2022 Practical study of communication in a multi-robot system.
MSL-RAPTOR: A 6DoF Relative Pose Tracker for Onboard Robotic Perception Benjamin Ramtoula^, Adam Caccavale^, Giovanni Beltrame, Mac Schwager ISER, 2020 An approach for estimating the 6DoF pose of objects from monocular images using hardware onboard a drone.
DOOR-SLAM: Distributed, Online, and Outlier Resilient SLAM for Robotic Teams Pierre-Yves Lajoie, Benjamin Ramtoula, Yun Chang, Luca Carlone, Giovanni Beltrame RA-L, 2020 video A fully distributed collaborative SLAM system with an outlier rejection mechanism.
CAPRICORN: Communication Aware Place Recognition using Interpretable Constellations of Objects in Robot Networks Benjamin Ramtoula^, Ricardo de Azambuja^, Giovanni Beltrame ICRA, 2020 An approach to do place recognition between multiple robots with minimal communication using a semantic representation of object layouts in scenes.
Planetary Exploration with Robot Teams: Implementing Higher Autonomy With Swarm Intelligence David St-Onge, Marcel Kaufmann, Jacopo Panerati, Benjamin Ramtoula, Yanjun Cao, Emily BJ Coffey, Giovanni Beltrame RAM, 2019 Description of our experiments as part of ESA's PANGAEA-X campaign, studying the cognitive load of astronauts to control fleets of robots in a Mars analog field experiment.