cv

Learning Sequential Kinematic Models from Demonstrations for Multi-Jointed Articulated Objects

Arxiv | Anmol Gupta, Weiwei Gu, Omkar Patil, Jun Ki Lee, Nakul Gopalan

Framework that learns kinematic constraints and manipulation sequences of multi-DoF objects from human demonstrations.

Factorizing Diffusion Policies for Observation Modality Prioritization

ICRA 2025 Workshop on Foundation Models and Neuro-Symbolic AI for Robotics| Omkar Patil, Prabin Rath, Kartikay Pangaonkar, Eric Rosen, Nakul Gopalan

Method to prioritize observational modalities, such as vision over tactile for learning diffusion policies.

Composing Diffusion Policies for Few-shot Learning of Motions

Compositional Learning Workshop @ NeurIPS 2024 | Omkar Patil, Anant Sah, Nakul Gopalan

Compositional approach that enables few-shot learning for novel skills by utilizing a combination of base policy priors is presented.

Learning Temporally Composable Task Segmentations with Language

IROS 2024 | Divyanshu Raj, Omkar Patil, Weiwei Gu, Chitta Baral and Nakul Gopalan

We present an approach to identify sub-tasks within a demonstrated robot trajectory with the supervision provided by language instructions.

Hardware-Software Co-Design for Path Planning by Drones

IROS 2024 | Ayushi Dube*, Omkar Patil*, Gian Singh, Nakul Gopalan, and Sarma Vrudhula

This work consists of designing a hardware-software co-design, MT+, for adapting the Mikami-Tabuchi (MT) algorithm for on-board path planning by drones in a 3D environment.

Understanding metrics for paraphrasing

Arxiv | Omkar Patil, Rahul Singh, Tarun Joshi

We propose a novel metric ROUGE-P to measure the quality of paraphrases along the dimensions of adequacy, novelty and fluency.

Document automation architectures and technologies: A survey

Arxiv | Mohammad Achachlouei*, Omkar Patil*, Tarun Joshi, Vijayan Nair

This paper surveys the current state of the art in document automation in light of recent advances in AI and deep neural networks.

Doctor of Philosophy

Arizona State University

Computer Science (3.9/4.0)

• Planning and Learning Methods for AI
• Knowledge Representation and Reasoning

Master of Technology

Indian Institute of Technology, Madras

Robotics and AI (8.22/10)

• Deep Learning
• Reinforcement Learning

Bachelor of Technology

Indian Institute of Technology, Madras

Mechanical Engineering (8.22/10)

• Computational Heat & Fluid Flow
• Design of Machine Elements

Graduate Intern

Robotics and AI Institute (Formerly BDAI)

Fall intern in the Compose research group, focusing on robot learning.

• Research at the intersection of performance improvement and steering for robot policies.
• Methodologies used include generative modeling (diffusion/flow) and reinforcement learning.

Graduate Research Assistant

Arizona State University

Research assistant in Dr. Nakul Gopalan's lab at Arizona State University.

• Developing robot learning algorithms that leverage compositionality for sample-efficient generalization.
• Developed a ROS2-based multi-drone formation and behavior control system in AirSim.

Senior NLP Quant

Wells Fargo

Member of the Artificial Intelligence and Automation within Corporate Model Risk.

• Researched and implemented prompt-tuning on language models for generating different kinds of paraphrases for downstream applications such as robustness testing.
• Collaborated with other researchers within the team to develop methodologies for evaluating model weaknesses with a special focus on text classification models.

NLP Quant

Wells Fargo

Member of the Artificial Intelligence and Automation within Corporate Model Risk.

• Explored text generation for the task of paraphrasing and developed a new metric to evaluate the quality of paraphrases.
• Surveyed various document automation frameworks present in literature.
• Contributed significantly to the internal code library and made several presentations on research projects, across the group

Head

Institute WebOps and MobOps

Lead of the official mobile development team of IIT Madras.

• Led a 9 member team for the development of the ‘Students App’, managing a budget of ~INR 3L.
• Increased the number of active users by ~160%, to 6500+ students, with 12000+ downloads in total.
• Developed a sophisticated Java front-end and PHP back-end to build a secure and useful application on Android.

Reviewer

Conferences

Reviewer for IROS 2024, ICRA 2025, CORL 2025

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