cv

My primary interest is in developing general purpose robots that can learn new skills on the fly. I believe compositional / modular learning and policy improvement will be crucial in enabling this.

Basics

Name Omkar Patil
Interests Robotics, NLP, Machine Learning
Email opatil3@asu.edu
Url https://omkarpatil18.github.io/
Summary I completed my Bachelor’s and Master’s degrees at IIT Madras, where I built a strong foundation in robotics and machine learning. I am currently pursuing my Ph.D. at Arizona State University (ASU) in Dr. Nakul Gopalan’s lab, focusing on robot learning, particularly methods that leverage compositionality to build generalizable manipulation skills.

Education

Publications

Work

  • 2025.07 - 2026.01

    Cambridge, MA

    Research Intern
    RAI Institute
    Post-training of VLAs and generative robot policies on downstream reward functions.
    • Developed a latent-steering-based approach that improved generative policies on 46 of 51 tasks without retraining, resulting in the filing of a patent and a publication: https://lottery-tickets.rai-inst.com/.
    • Demonstrated that derivative-free episodic policy improvement can be competitive with policy-gradient and actor-critic methods, especially for VLAs, potentially with a 15× reduction in wall-clock time.
  • 2023.08 - Present

    Tempe, Arizona

    Graduate Research Assistant
    Logos Robotics Lab, Arizona State University
    PhD research on robot learning for real-world manipulation with Dr. Nakul Gopalan.
    • Build ROS 2 teleoperation pipelines (leader arm, VR headset) for a Franka Research 3 arm and a ROBOTIS AI Worker semi-humanoid, used to collect 500+ demonstrations across 10 tasks.
    • Build data-collection and real-time inference pipelines to deploy trained policies, including Diffusion Policy, ACT, and VLAs (π0.5, SmolVLA), on real robots.
    • Fine-tune policies with behavior cloning and on-robot policy improvement, and deploy simulation-trained policies on hardware (sim-to-real) for research experiments and demonstrations.
  • 2020.08 - 2023.07

    Bengaluru, India

    Senior NLP Quantitative Researcher
    Wells Fargo
    • Developed a novel metric to evaluate the quality of LLM-generated paraphrases (arXiv, patent).
    • Surveyed and built LLM-enabled document generation, speeding up internal documentation (arXiv).

Volunteer

  • 2026.10 - Present
    Organizer
    Compositional and Modular Learning in the Era of Scaling in Robotics (IROS 2026)
    Workshop on compositional and modular robot learning at IROS 2026.
  • 2025.01 - Present
    Reviewer
    Conferences
    CoRL 2025, 2026; ICRA 2025, 2026; IROS 2025, 2026
  • 2019.11 - 2019.12
    Intern
    Madhuvan Foundation
    Non-governmental organization for the welfare of underprivileged children.
    • Quickly ramped up with the current operations of reselling scrap newspaper and facilitating education for the needy with the obtained proceeds
    • Streamlined the scrap collection under the ‘Pasti Ki Pathshala' project by creating a database of participants willing to donate scrap newspaper
    • Organized a collection drive at a local school for 800+ students to promote the idea of social welfare

Skills

Robot Learning
Imitation Learning
Behavior Cloning
VLA Post-training (Latent Steering, Policy Improvement)
Generative Policies (Diffusion, Flow Matching, ACT)
Reinforcement Learning (PPO, SAC, Offline RL, On-robot RL, Sim-to-Real)
Robotics
Motion Planning (IK, Trajectory Optimization), Perception (Vision, Point Clouds), Control
Hardware (Franka FR3, ROBOTIS AI Worker), Teleoperation (Leader Arm, VR), ROS 2, MuJoCo, Isaac Sim
Tooling
VLAs (π0.5, SmolVLA, GR00T N1.5), LeRobot, PyTorch, JAX, Hugging Face, W&B, Docker, Git