Vidur Sanghi

I am an undergraduate student at Carnegie Mellon University pursuing a degree in Computer Science. I am a researcher at the roBot Intelligence Group (BIG) and the Momentum Lab in the Robotics Institute, where I am advised by Uksang Yoo, Jeff Ichnowski, and Jean Oh.

I will soon be joining ETH Zürich as a Visiting Research Assistant at the Computer Vision and Geometry (CVG) Lab, where I will work with Yung-Hsu (Roy) Yang under the supervision of Prof. Marc Pollefeys. My research there will focus on scene understanding and 3D object detection and tracking.

My research interests are in soft robotics, reinforcement learning, computer vision, and developing AI systems that can work reliably in real-world scenarios. I'm particularly focused on bridging the gap between cutting-edge AI methods and practical applications where that technology can make a meaningful impact but doesn't yet work effectively. My research is generously supported by the BXA Grant Program and CMU SURA.

Currently, I'm applying reinforcement learning to train soft robotic hands to perform delicate manipulation tasks. I'm working with John Francis from Bosch AI to develop an API controller for open-source research and make it development-ready. I presented my findings at the IEEE ICRA 2025 soft robotics reproducibility workshop in Atlanta, Georgia, and I'm currently working towards an IEEE T-RO publication.

I'm also an advisor and developer at Ease Vertical AI, where we're building AI integration tools for small businesses with a focus on Go-To-Market strategy and lead generation. Additionally, I have joined Atomic VC, a venture capital studio, as a Machine Learning Intern where I advise on investments and apply ML to help their portfolio companies, advised by Nikita Elkin.