Vaidehi Wagh
Master of Science in Robotics, Carnegie Mellon University

I am a graduate researcher at Carnegie Mellon University, where my research focuses on building multimodal interfaces for human-exoskeleton interaction, building end-to-end, latency, resource and safety aware systems that can adapt to users through life.

I received my B.Tech. in Mechanical Engineering from the College of Engineering, Pune, and was a research intern at the University of British Columbia (Neuroplasticity, Imagery and Motor Behaviour Laboratory), where I built large-scale vision pipelines for upper-limb kinematics after stroke.

I am seeking full-time research positions in machine learning, artificial intelligence, and robotics research roles with availability from June 2026.

Research

Preview still for exoskeleton personalization research
Exoskeleton control personalization MetaMobility Lab, Carnegie Mellon University

Multimodal, user-in-the-loop hip assistance: egocentric video and transcribed speech inform vision-language reasoning and structured controller updates for wearable deployment.

Preview still for stroke kinematics and MediaPipe research
Clinical kinematic analysis with monocular pose estimation Neuroplasticity, Imagery and Motor Behaviour Lab, University of British Columbia

High-throughput analysis of post-stroke upper-limb reaching using commodity vision, with validation against laboratory motion capture and peer-reviewed clinical reporting.

Publications

Transfer learning for biological joint moment estimation in stroke populations
Wagh V., Park D., Young A., Kang I.
American Society of Biomechanics, 2025
Using MediaPipe to track upper-limb reaching movements after stroke: a proof-of-principle study
Wagh V., Scott M., Andrushko J., Jones C., Larrsen B., Boyd L., Kraeutner S.
Journal of Neuroengineering and Rehabilitation, 2025
Quantifying Similarities Between MediaPipe and a Known Standard to Address Issues in Tracking 2D Upper Limb Trajectories: Proof of Concept Study
Wagh V., Scott M., Kraeutner S.
Journal of Medical Internet Research, Formative Research, 2024

Projects

Preview for multimodal human–robot interface course project
Vision-language interface for human–robot interaction Course project, Talking to Robots (Fall 2025)

Finetuned compact VLMs for speech- and vision-conditioned planning relevant to personalized assistive control.

Preview for automated IV insertion course prototype
Automated intravenous insertion prototype Course project, Medical Robotics (Fall 2025)

Custom actuation, vision-based vein localization, and closed-loop integration for a bench-scale insertion workflow.

Preview for imitation learning course project
Generative adversarial imitation learning Course project, Introduction to Robot Learning (Spring 2025)

Locomotion imitation from motion-capture experts using adversarial reward inference and policy optimization.

Preview for LQR humanoid balancing course project
LQR-based humanoid balancing Course project, Optimal Control and Reinforcement Learning (Spring 2025)

Linear-quadratic regulation for balance of a simulated humanoid on a moving support surface.

Teaching

11977A: Multimodal Machine Learning: S26 Prof. Yonatan Bisk, Prof. David Mortensen, Prof. Ralf Brown
16385: Computer Vision: F25 Prof. David Held
PLUS Tutoring: S25 - S26 Lead Tutor