Antoine Deleforge

About

Antoine Deleforge is a researcher whose work sits at the dynamic intersection of machine learning, audio signal processing, and human-robot interaction. His research spans sound source localization and separation, head-pose estimation, and multimodal perception systems — areas with profound implications for how robots and computers understand the world around them. Deleforge's most influential contribution, "Robust Head-Pose Estimation Based on Partially-Latent Mixture of Linear Regressions" (2017, 80 citations), introduced sophisticated probabilistic methods for estimating human head orientation under real-world variabilities — a critical capability for applications ranging from autonomous driving to social robotics. His pioneering work on binaural sound-source localization explored the striking finding that high-dimensional auditory data lies on a locally-linear manifold, enabling robots to learn spatial hearing in a manner inspired by human perception. A recurring theme throughout his research is giving robots genuinely human-like perceptual abilities. His cocktail party robot work, active-speaker detection systems, and ego-noise reduction algorithms all tackle the messy acoustic realities robots face in real environments. The RAVEL corpus further demonstrates his commitment to building resources that advance the entire robotics community. With over 300 cumulative citations, Deleforge has established himself as an important voice in robot audition and audiovisual perception research.

Research Focus

Key Achievements

Top Papers

1. 1
   Robust Head-Pose Estimation Based on Partially-Latent Mixture of Linear Regressions

   80 citations · 2017
2. 2
   2D sound-source localization on the binaural manifold

   43 citations · 2012
3. 3
   Active-speaker detection and localization with microphones and cameras embedded into a robotic head

   39 citations · 2013
4. 4
   RAVEL: an annotated corpus for training robots with audiovisual abilities

   35 citations · 2012
5. 5
   The cocktail party robot

   31 citations · 2012
6. 6
   Phase-optimized K-SVD for signal extraction from underdetermined multichannel sparse mixtures

   25 citations · 2015
7. 7
   Challenges in Acoustic Signal Enhancement for Human-Robot Communication

   22 citations · 2014
8. 8
   Online multimodal speaker detection for humanoid robots

   16 citations · 2012
9. 9
   Ego-noise reduction using a motor data-guided multichannel dictionary

   15 citations · 2016
10. 10
    Acoustic Space Mapping : A Machine Learning Approach to Sound Source Separation and Localization

    8 citations · 2013

Key Collaborators

Contact & Links