Akshara Rai

About

Akshara Rai is a robotics researcher whose work sits at the intersection of reinforcement learning, locomotion, and robot manipulation. Her research spans a remarkably broad terrain — from teaching bipedal and quadrupedal robots to walk and navigate complex environments, to enabling robotic arms to assist elderly and disabled individuals with everyday tasks like dressing. Early work on robotic clothing assistance (100 citations) demonstrated her commitment to socially impactful robotics, while subsequent contributions to deep reinforcement learning for bipedal control and hierarchical locomotion frameworks have shaped how the field approaches real-world robot deployment. Rai has made notable strides in sample-efficient learning, developing hierarchical reinforcement learning strategies and latent action spaces that allow legged robots to generalize skills across diverse terrains. Her research extends into graph neural networks for interpretable manipulation and Bayesian optimization for high-dimensional parameter tuning. More recently, she has tackled long-horizon mobile manipulation and multi-robot cable-towing coordination, reflecting a systems-level vision for autonomous robotics. With over 500 cumulative citations across a decade of work, Rai has established herself as a versatile and influential voice in modern robot learning research.

Research Focus

Key Achievements

Top Papers

1. 1
   Reinforcement learning of clothing assistance with a dual-arm robot

   100 citations · 2011
2. 2
   Using Deep Reinforcement Learning to Learn High-Level Policies on the ATRIAS Biped

   58 citations · 2019
3. 3
   Learning coupling terms for obstacle avoidance

   47 citations · 2014
4. 4
   Efficient and Interpretable Robot Manipulation With Graph Neural Networks

   44 citations · 2022
5. 5
   Learning Generalizable Locomotion Skills with Hierarchical Reinforcement Learning

   42 citations · 2020
6. 6
   Learning Torque Control for Quadrupedal Locomotion

   39 citations · 2023
7. 7
   ASC: Adaptive Skill Coordination for Robotic Mobile Manipulation

   36 citations · 2023
8. 8
   Collaborative Navigation and Manipulation of a Cable-Towed Load by Multiple Quadrupedal Robots

   35 citations · 2022
9. 9
   Re-Examining Linear Embeddings for High-Dimensional Bayesian Optimization

   32 citations · 2020
10. 10
    Planning in Learned Latent Action Spaces for Generalizable Legged Locomotion

    28 citations · 2021

Key Collaborators

Contact & Links