Fangyi Zhang

About

Fangyi Zhang is a robotics and machine learning researcher whose work sits at the intersection of computer vision, deep reinforcement learning, and autonomous robot control. He is perhaps best known for his pioneering contributions to sim-to-real transfer — the challenge of training robotic agents in simulation and deploying them effectively in the real world. His adversarial discriminative sim-to-real transfer framework (2019, 40 citations) addresses a critical bottleneck by reducing dependence on labeled real-world data, while his earlier modular Deep Q-Network approach (2016, 33 citations) laid foundational groundwork for practical visuo-motor policy learning on physical robots. Zhang also made significant contributions to reproducible robotics research through the ACRV Picking Benchmark (2017, 81 citations), which established a standardized platform for evaluating robotic shelf-picking systems — his most-cited work to date. Earlier in his career, he explored indoor localization using Visible Light Communication combined with Gaussian Processes (2015, 31 citations), demonstrating versatility across sensing modalities. Additional contributions include fine-tuning strategies for modular networks and metric-free path planning for mobile robots. Across his body of work, Zhang consistently bridges the gap between controlled experimental settings and real-world robotic deployment, making his research highly relevant to the broader autonomous systems community.

Research Focus

Key Achievements

Top Papers

1. 1
   The ACRV picking benchmark: A robotic shelf picking benchmark to foster reproducible research

   81 citations · 2017
2. 2
   Adversarial discriminative sim-to-real transfer of visuo-motor policies

   40 citations · 2019
3. 3
   Modular Deep Q Networks for Sim-to-real Transfer of Visuo-motor Policies

   33 citations · 2016
4. 4
   Visible Light Communication-based indoor localization using Gaussian Process

   31 citations · 2015
5. 5
   Adversarial Discriminative Sim-to-real Transfer of Visuo-motor Policies

   10 citations · 2017
6. 6
   Asynchronous blind signal decomposition using tiny-length code for Visible Light Communication-based indoor localization

   8 citations · 2015
7. 7
   Visible light communication-based indoor environment modeling and metric-free path planning

   5 citations · 2015
8. 8
   The ACRV Picking Benchmark (APB): A Robotic Shelf Picking Benchmark to Foster Reproducible Research

   5 citations · 2016
9. 9
   Tuning Modular Networks with Weighted Losses for Hand-Eye Coordination

   4 citations · 2017
10. 10
    Transferring Vision-based Robotic Reaching Skills from Simulation to Real World

    3 citations · 2016

Key Collaborators

Contact & Links