Ajay U. Mandlekar

Picture of Ajay. 

Ajay Mandlekar
Ph.D. Candidate
Stanford Vision and Learning Lab
Department of Electrical Engineering
Stanford University

Advisors: Silvio Savarese and Fei-Fei Li


Email: amandlek@stanford.edu

Recent News

  • We released our paper on offline learning from large-scale robot manipulation datasets and demonstrated that it can learn from our crowdsourced RoboTurk dataset without requiring additional interaction. See our paper, website, and video for more details!

  • See our paper, blog post, and video on learning latent action spaces for assisted teleoperation!

  • Our paper on leveraging an ensemble of partial solutions to guide agent exploration during reinforcement learning appeared at CoRL 2019. Check out our blog post, website, and talk for more details!

About Me

I am a PhD candidate in the Stanford Vision and Learning Lab advised by Silvio Savarese and Fei-Fei Li. I work on applying Reinforcement Learning and Imitation Learning to Robotics. In particular, I'm interested in building systems and algorithms to allow robots to leverage human insight for manipulation.

Currently, I am developing a data collection system for robotic manipulation tasks. The core goal of our system is to enable large-scale human supervision for robots.

This data can be used in algorithms that combine Imitation and Reinforcement Learning to both speed up policy learning and also enable learning on tasks that could not be solved with Reinforcement Learning alone.