Alexandre Alahi

About

Alexandre Alahi is a leading researcher at the intersection of computer vision, human motion understanding, and socially intelligent robotics. His work centers on three interconnected domains: multi-object tracking, trajectory prediction, and social robot navigation — all united by the goal of enabling machines to perceive and navigate human-centric environments safely and naturally. Alahi's early breakthrough came with his 2015 work on online multi-object tracking through decision-making, which has garnered over 716 citations and became a foundational reference in the tracking-by-detection paradigm. He further advanced the field with Social GAN (2018), applying generative adversarial networks to model the inherently multimodal nature of human motion, and later introduced Social NCE, leveraging contrastive learning to improve generalization in socially-aware motion representations. Perhaps most impactful is his research on crowd-aware robot navigation using attention-based deep reinforcement learning, cited over 581 times, which demonstrated how robots can learn socially compliant behavior even in dense crowds. His PifPaf framework for human pose estimation and his community guidelines for evaluating social navigation algorithms further reflect his commitment to rigorous, deployable AI. Alahi's body of work has become essential reading for researchers building autonomous systems that must coexist intelligently with people.

Research Focus

Key Achievements

Top Papers

1. 1
   Learning to Track: Online Multi-object Tracking by Decision Making

   716 citations · 2015
2. 2
   Crowd-Robot Interaction: Crowd-Aware Robot Navigation With Attention-Based Deep Reinforcement Learning

   581 citations · 2019
3. 3
   Social GAN: Socially Acceptable Trajectories with Generative Adversarial Networks

   180 citations · 2018
4. 4
   Social NCE: Contrastive Learning of Socially-aware Motion Representations

   112 citations · 2021
5. 5
   Principles and Guidelines for Evaluating Social Robot Navigation Algorithms

   59 citations · 2024
6. 6
   Long-term path prediction in urban scenarios using circular distributions

   45 citations · 2017
7. 7
   OpenPifPaf: Composite Fields for Semantic Keypoint Detection and Spatio-Temporal Association

   22 citations · 2021
8. 8
   PifPaf: Composite Fields for Human Pose Estimation

   20 citations · 2019
9. 9
   Principles and Guidelines for Evaluating Social Robot Navigation Algorithms

   15 citations · 2023
10. 10
    Crowd-Robot Interaction: Crowd-aware Robot Navigation with Attention-based Deep Reinforcement Learning

    12 citations · 2018

Key Collaborators

Contact & Links