yuqing_du (at) berkeley (dot) edu
Hello! I am a fourth year PhD student at UC Berkeley advised by Professor Pieter Abbeel at the Berkeley Artificial Intelligence Research (BAIR) Lab. I received my B.A.Sc in Engineering Physics from the University of British Columbia, where I was advised by Professors Machiel Van der Loos and Elizabeth Croft at the Collaborative Advanced Robotics and Intelligent Systems (CARIS) Lab.
Currently I am interested in developing unsupervised reinforcement learning methods to reduce reliance on hand-crafted rewards and highly curated datasets. Some directions I am curious about: 1) understanding which objectives can successfully intrinsically motivate agents to learn emergent, useful behaviours, 2) how can we leverage human demonstrations, priors, and feedback to bootstrap exploration in open-ended environments, and 3) how can we use multi-agent interactions to automatically generate learning curricula. Specific applications I am interested in include real-world robotics and assistive software agents.
- ICML 2022Bayesian Imitation Learning for End-to-End Mobile ManipulationInternational Conference on Machine Learning (ICML) 2022
- ICLR 2022It Takes Four to Tango: Multiagent Selfplay for Automatic Curriculum GenerationInternational Conference on Learning Representations (ICLR) 2022
- ICRA 2021Auto-Tuned Sim-to-Real Transfer
(Best Cognitive Robotics Paper Finalist)2021 International Conference on Robotics and Automation (ICRA)
- NeurIPS 2020AvE: Assistance via Empowerment2020 Neural Information Processing Systems (NeurIPS)
- ICRA 2019Group Surfing: A Pedestrian-Based Approach to Sidewalk Robot Navigation2019 International Conference on Robotics and Automation (ICRA)
- Practical Imitation Learning in the Real World via Task Consistency Loss
- Robust Reinforcement Learning using Adversarial Populations