REACH Lab — Robot Learning and Control Lab

The Robot Learning and Control Lab at NYU Abu Dhabi develops methods that let robots learn, adapt, and act safely in the physical world. Our work sits between control, which carries guarantees but needs a model, and learning, which brings flexibility but offers no promises. The gap between them opens wherever contact matters, and that is where we work — across machine learning, control theory, differential geometry, and dexterous manipulation.

Our mission is to build robots that acquire skills from demonstration and from their own interaction, manipulate dexterously, work alongside people, and make sound decisions under uncertainty — on rigid manipulators, on soft and continuum robots, and in assistive devices. We draw on reinforcement learning, learning from demonstration, model predictive control, and certified safe learning, and we test what we build on real hardware.

Five Pillars

Five pillars carry the work, and give the lab its name:

  1. Riemannian and geometric robot learning
  2. Embodied skill learning, from demonstration and interaction
  3. Adaptive and compliant manipulation
  4. Certifiable learning-based control
  5. Human-centered and assistive robotics

We teach robots not just to move, but to understand, adapt, and collaborate safely in a world built for humans.

Fares Abu-Dakka, Principal Investigator

Contact Us

Principal Investigator

Fares Abu-Dakka
Assistant Professor of Mechanical Engineering, Assistant Professor of Electrical Engineering (Affiliated), NYU Abu Dhabi; Associated Faculty with the Mechanical Engineering Department, Associated Faculty with the Electrical and Computer Engineering Department, NYU Tandon, NYU Tandon PhD Mentor
Email: fa2656@nyu.edu