Abdeslam Boularias
Carnegie Mellon University, Rutgers, The State University of New Jersey, Max Planck Institute for Intelligent Systems, Max Planck Society, Max Planck Institute for Biological Cybernetics, Rutgers Sexual and Reproductive Health and Rights, Laboratoire d'Informatique de Paris-Nord, Université Laval
Papers
60
Total Citations
1,047
H-Index
18
About
Abdeslam Boularias is a robotics and machine learning researcher whose work sits at the intersection of autonomous manipulation, reinforcement learning, and perception. His research has made substantial contributions to some of the most challenging problems in modern robotics: enabling robots to grasp and manipulate unknown objects in cluttered environments, learn from physical interaction, and interpret natural language commands for navigation. Boularias pioneered approaches that allow robots to learn entirely from scratch through trial and error, eliminating dependence on pre-built object models — a breakthrough demonstrated in his widely cited 2015 work on manipulation in clutter (103 citations). His 2017 self-supervised object detection system cleverly leveraged physics simulation and multi-view pose estimation to sidestep the burdensome data-labeling process that plagues deep learning applications in robotics (100 citations). His Deep Interaction Prediction Network (DIPN) further advanced the field by enabling robots to mentally "imagine" the consequences of pushing actions before execution. Beyond manipulation, Boularias has explored inverse reinforcement learning in dynamic settings like table tennis and developed compliant, vision-driven strategies for high-precision assembly. With a body of work spanning probabilistic learning, physics-based modeling, and language-grounded navigation, his research consistently pushes robots toward greater autonomy and adaptability in real-world environments.
Research Focus
Key Achievements
Top Papers
- 1Learning to Manipulate Unknown Objects in Clutter by Reinforcement103 citations · 2015
- 2
- 3Learning strategies in table tennis using inverse reinforcement learning70 citations · 2014
- 4
- 5
- 6Grounding spatial relations for outdoor robot navigation44 citations · 2015
- 7
- 8
- 9Anticipatory action selection for human–robot table tennis40 citations · 2014
- 10Learning robot grasping from 3-D images with Markov Random Fields40 citations · 2011