Anukriti Singh

I am a Ph.D. student in Computer Science at the University of Maryland, College Park, advised by Prof. Pratap Tokekar in the RAAS Lab. I hold a Master of Engineering (Robotics) from UMD.

My research sits at the intersection of perception and robot learning, with a focus on building manipulation and navigation systems that generalize in the real world. I'm especially interested in learning from rich supervision — video, language, demonstrations, and preference feedback — and using foundation-scale vision models to make policies and rewards more data-efficient and robust.

Email / CV / Scholar / LinkedIn / GitHub / X

News

Jul 2026 STeP paper posted on arXiv — precise action generation with Vision Language Models.
May 2026 Moved to Paris for the summer as a Research Intern at Enchanted Tools!
Jan 2025 Sketch-to-Skill accepted to RSS 2025!

Jul 2025 First time volunteering at RSS 2025.
Mar 2025 Awarded NSF I-Corps 2025 grant for customer discovery in tech innovation.
Jan 2025 Starting my Ph.D. at UMD with Prof. Pratap Tokekar!
Mar 2024 Won the ICRA 2024 Travel Grant to attend ICRA in Yokohama, Japan.
Feb 2024 Joining Bell Labs as a Spring co-op research intern!
Jan 2024 Pre-Trained Masked Image Model for Navigation accepted to ICRA 2024!
Aug 2023 SeMask accepted to NIVT Workshop at ICCV 2023!

Experience

May – Sep 2026 Paris, France

Research Intern · Enchanted Tools

Researched methods to improve policy performance using lightweight reinforcement learning and deployed algorithms on the Mirokai humanoid robot for mobile manipulation.

Lightweight RL Mirokai Robot Mobile Manipulation
Jun – Dec 2024 College Park, MD

Faculty Research Assistant · UMIACS, UMD

Developed a framework for bootstrapping RL using 2D sketch trajectories. Achieved 96% of teleop baseline and 170% improvement over pure RL.

Reinforcement Learning Trajectory Sketching
📄 → Sketch-to-Skill (RSS 2025)
Feb – May 2024 NJ / Remote

Robotics Co-op · Nokia Bell Labs

Curated large-scale warehouse data for segmentation models. Adapted 3D multi-object tracking to 2D using Kalman filters.

Kalman Filtering 3D MOT

Jun – Aug 2023 Murray Hill, NJ

Robotics Intern · Nokia Bell Labs

Engineered a zero-shot 6D pose estimation pipeline (YOLO + SfM + PnP). Optimized efficiency by 50%.

6D Pose Estimation Zero-Shot

Research

I'm interested in robot manipulation, perception, and learning from rich supervision. Representative papers are highlighted.

	STeP: Signal Temporal Logic for Precise Specifications for Action Generation with Vision Language Models K. Torshizi, Anukriti Singh, S. Mathur, P. Tokekar In Submission arXiv, 2026 PDF
	Afford2Act: Affordance-Guided Automatic Keypoint Selection for Generalizable and Lightweight Robotic Manipulation Anukriti Singh, K. Torshizi, K. Habib, K. Yu, R. Gao, P. Tokekar In Submission PDF
	VARP: Reinforcement Learning from Vision-Language Model Feedback with Agent Regularized Preferences Anukriti Singh, A. Bhaskar, P. Yu, S. Chakraborty, R. Dasyam, A. Bedi, P. Tokekar In Submission PDF
	Sketch-to-Skill: Bootstrapping Robot Learning with Human Drawn Trajectory Sketches P. Yu, A. Bhaskar, Anukriti Singh, Z. Mahammad, P. Tokekar RSS 2025 PDF
	Pre-Trained Masked Image Model for Mobile Robot Navigation V. D. Sharma, Anukriti Singh, P. Tokekar ICRA, 2024 PDF
	FLIP-TD: Free Lunch Inpainting on Top-Down Images for Robotic Tasks Anukriti Singh, V. D. Sharma, P. Tokekar ICRA Workshop, 2023 PDF
	SeMask: Semantically Masked Transformers for Semantic Segmentation J. Jain, Anukriti Singh, N. Orlov, Z. Huang, J. Li, S. Walton, H. Shi ICCV Workshop, 2023 PDF
	Characterizing Quantum Evolutions via a Recommender System P. Batra, Anukriti Singh, T. S. Mahesh Quantum Journal, 2021 PDF