Babak Nadjar Araabi

About

Babak Nadjar Araabi is a prominent researcher whose work sits at the intersection of machine learning, human-robot interaction, and computational intelligence. His research spans visual attention modeling, imitation learning, reinforcement learning, and concept-based knowledge representation — areas that collectively address how robots and intelligent systems can learn naturally and efficiently from human guidance. Araabi's most cited contribution, "Online learning of task-driven object-based visual attention control" (2009, 69 citations), demonstrated how robots could develop adaptive, context-aware visual focus, with complementary work on saliency-based hand gesture recognition advancing real-time human-robot interaction. A defining thread throughout his career is making robot learning more robust under realistic, imperfect conditions — his research on inverse reinforcement learning tackles the fundamental challenge that human demonstrations are often sparse and inaccurate, enabling systems to still extract meaningful behavior from flawed teachers. His sustained exploration of conceptual imitation learning is particularly notable, moving beyond simple mimicry toward genuine abstraction — teaching robots to understand the *meaning* behind human actions rather than merely copying their form. Combined with his work on knowledge-based cooperation between learning agents, Araabi has made substantial contributions to building intelligent systems capable of natural, human-like learning in complex real-world environments.

Research Focus

Key Achievements

Top Papers

1. 1
   Online learning of task-driven object-based visual attention control

   69 citations · 2009
2. 2
   Learning from Demonstrations and Human Evaluative Feedbacks: Handling Sparsity and Imperfection Using Inverse Reinforcement Learning Approach

   20 citations · 2020
3. 3
   Fast Hand gesture recognition based on saliency maps: An application to interactive robotic marionette playing

   17 citations · 2009
4. 4
   Design, implementation and analysis of an alternation-based Central Pattern Generator for multidimensional trajectory generation

   17 citations · 2011
5. 5
   Knowledge-based Extraction of Area of Expertise for Cooperation in Learning

   16 citations · 2006
6. 6
   Combination of learning from non-optimal demonstrations and feedbacks using inverse reinforcement learning and Bayesian policy improvement

   12 citations · 2018
7. 7
   Concept Oriented Imitation Towards Verbal Human-Robot Interaction

   11 citations · 2006
8. 8
   Conceptual Imitation Learning Based on Perceptual and Functional Characteristics of Action

   9 citations · 2013
9. 9
   Conceptual Imitation Learning in a Human-Robot Interaction Paradigm

   8 citations · 2012
10. 10
    A Probabilistic Reinforcement-Based Approach To Conceptualization

    8 citations · 2008

Key Collaborators

Contact & Links