Yunlong Song <宋运龙>


Google Scholar, LinkedIn
About

Yunlong (运龙, the cloud-walking dragon in chinese) Song is a PhD student at the Robotics and Perception Group led by Prof. Davide Scaramuzza at the University of Zurich and ETH Zurich. Also, he is an Associated PhD Student at ETH AI Center. Prior to RPG, he obtained his Msc degree in Information and Communication Engineering (iCE) from Technische Universität Darmstadt. His master thesis entitled "Minimax and Entropic Proximal Policy Optimization" was written under the supervision of Boris Belousov and Prof. Jan Peters. Yunlong was born and grew up in a small (probably one of the most isolated) village in Southern China.



Research Highlights

Yunlong's research interests lie at the intersection of Reinforcement Learning (RL) and Optimal Control (OC) with a focus on real-world applications. His Ph.D. topic is Reinforcement Learning for Agile Flight.

Reinforcement Learning and Optimal Control

  • Policy Search for Model Predictive Control with Application to Agile Drone Flight (T-RO, 2022).
  • Learning High-Level Policies for Model Predictive Control (IROS, 2021).
  • Reinforcement Learning Algorithms: Analysis and Applications (chapter Information-Loss-Bounded Policy Optimization) (Springer Cham, 2021)
  • Robot Learning Applications

  • Learning Minimum-Time Flight in Cluttered Environments (RA-L + IROS 2022).
  • Autonomous Drone Racing with Deep Reinforcement Learning (IROS 2021).
  • Super-Human Performance in Gran Turismo Sport Using Deep Reinforcement Learning (RA-L + ICRA 2021).
  • Autonomous Overtaking in Gran Turismo Sport Using Curriculum Reinforcement Learning (ICRA 2021).
  • Simulation and Tools

  • Agilicious: Open-source and Open-hardware Agile Quadrotor for Vision-based Flight (Science Robotics, 2022).
  • Flightmare: A Flexible Quadrotor simulator (CoRL 2020).
  • Submitted

  • Learning Perception-aware Agile Flight in Cluttered Environments (under review).
  • Maximum Likelihood for Model Predictive Contouring Control (under review).
  • Learning Deep Sensorimotor Policies for Vision-based Autonomous Drone Racing (under review).

  • Open-source Projects
  • Flightmare Star
  • Agilicious Star
  • High-MPC Star
  • DodgeDrone Challenge Star

  • Contact

    Yunlong Song
    University of Zurich
    Department of Informatics
    Robotics and Perception Group
    Andreasstrasse 15, AND 2.08
    8050 Zurich
    Switzerland

    Email: song (at) ifi (dot) uzh (dot) ch
    Office: +41 44 635 45 89