Research
I am generally interested in deep learning, representation learning, computer vision, and their application to robotics. I have been working on foundation models, focusing on how they can be leveraged for tasks in robotics, such as embodied perception, tactile sensing, and manipulation.
Your browser does not support the video tag.
Sensor-Invariant Tactile Represenation
Harsh Gupta* ,
Yuchen Mo* ,
Shengmiao Jin ,
Wenzhen Yuan
International Conference on Learning Representations (ICLR) , 2025
arXiv /
webpage
Tactile foundation model enabling zero-shot transfer across optical tactile sensors by extracting sensor-invariant tactile representations.
Projects
Here are some fun projects I worked on before I knew what research was.
Your browser does not support the video tag.
Newton Hand: Real-Time Spatial Modeling and Gesture Estimation
Fall 2023
video /
code
A wearable device with IMUs for real-time hand landmark estimation, enabling basic American Sign Language (ASL) translation.
Your browser does not support the video tag.
TRACE: Tennis Match Analysis and Visualization Tool
Spring 2023
video /
code
A tool that processes tennis match videos to extract player, court, and ball data, generating top-down views and collecting data for analysis.