Chen Yu

MPhil Student in Robotics & AI


I am a master's student at ShanghaiTech University working on robot learning, supervised by Prof. Andre Rosendo and Prof. Laurent Kneip. Before that, I obtained three Bachelor's degrees from University of Glasgow (First-class Honour) and University of Electronic Science and Technology of China in 2020. My research focuses on creating sample-efficient Reinforcement Learning or Bayesian Optimisation algorithms that endow legged robots with adaptive skills for challenging locomotion scenarios or agile manoeuvres. I'll be joining Northwestern University's Xenobot Lab this fall!


Multi-embodiment Legged Robot Control as a Sequence Modeling Problem

Chen Yu, Weinan Zhang, Hang Lai, Zheng Tian, Laurent Kneip, and Jun Wang
IEEE International Conference on Robotics and Automation (ICRA) 2023
(in press)


Co-optimizing control and morphology of a robot is often inefficient due to the complex interplay between the controller and morphology. In this paper, we propose a learning-based control method that can inherently take morphology into consideration such that once the control policy is trained in the simulator, it can be easily deployed to robots with different embodiments in the real world. In particular, we present the Embodiment-aware Transformer (EAT), an architecture that casts this control problem as conditional sequence modeling. Experimental results show that EAT can outperform all other alternatives in embodiment-varying tasks, and succeed in an example of real-world evolution tasks: stepping down a stair through updating the morphology alone.

Multi-Modal Legged Locomotion Framework with Automated Residual Reinforcement Learning

Chen Yu and Andre Rosendo
IEEE Robotics and Automation Letters (RA-L) 2022
IEEE International Conference on Intelligent Robots and Systems (IROS) 2022

Webpage    IEEE Xplore    Video

While quadruped robots usually have good stability and load capacity, bipedal robots offer a higher level of flexibility / adaptability to different tasks and environments. A multi-modal legged robot can take the best of both worlds. In this paper, we propose a multi-modal locomotion framework that is composed of a hand-crafted transition motion and a learning-based bipedal controller—learnt by a novel algorithm called Automated Residual Reinforcement Learning. This framework aims to endow arbitrary quadruped robots with the ability to walk bipedally. Overall, experimental results show that our multi-modal robot could successfully switch between biped and quadruped, and walk in both modes.

Learning to Climb: Constrained Contextual Bayesian Optimisation on a Multi-Modal Legged Robot

Chen Yu, Jinyue Cao and Andre Rosendo
IEEE Robotics and Automation Letters (RA-L) 2022
IEEE International Conference on Intelligent Robots and Systems (IROS) 2022

Webpage    IEEE Xplore    Video

Controlling a legged robot to climb obstacles with different heights is challenging, but important for an autonomous robot to work in an unstructured environment. In this paper, we model this problem as a novel contextual constrained multi-armed bandit framework. We further propose a learning-based Constrained Contextual Bayesian Optimisation (CoCoBo) algorithm that can solve this class of problems efficiently. CoCoBo models both the reward function and constraints as Gaussian processes, incorporate continuous context space and action space into each Gaussian process, and find the next training samples through excursion search. The experimental results show that CoCoBo is more data-efficient and safe, compared to other related state-of-the-art optimisation methods, on both synthetic test functions and real-world experiments.

Bridging the Reality Gap via Progressive Bayesian Optimisation

Chen Yu and Andre Rosendo
International Conference Series on Climbing and Walking Robots and the Support Technologies for Mobile Machines (CLAWAR) 2022


The recent rapid development of deep-learning-based control strategies has made the reality gap a critical issue at the forefront of robotics, especially for legged robots. We propose a novel system identification framework, Progressive Bayesian Optimisation (ProBO), to bridge the reality gap by tuning simulation parameters. Since dynamic locomotion trajectories are usually harder to narrow the reality gap than their static counterpart, we train a Gaussian process model with the easier trajectory data set and make it a prior to start the learning process of a harder one. We implement ProBO on a quadruped robot to narrow the reality gaps of a set of bounding gaits at different speeds. Results show that our methods can outperform all other alternatives after training the initial gait.

Risk-aware Model-based Control

Chen Yu and Andre Rosendo
Frontiers of Robotics and AI 2021

HTML    Code

Model-Based Reinforcement Learning (MBRL) algorithms have been shown to have an advantage on data-efficiency, but often overshadowed by state-of-the-art model-free methods in performance, especially when facing high-dimensional and complex problems. In this work, a novel MBRL method is proposed, called Risk-Aware Model-Based Control (RAMCO). It combines uncertainty-aware deep dynamics models and the risk assessment technique Conditional Value at Risk (CVaR). This mechanism is appropriate for real-world application since it takes epistemic risk into consideration. In comparison with other state-of-the-art reinforcement learning algorithms, we show that it produces superior results on a walking robot model.

Sim-to-Real Transfer for Quadrupedal Locomotion via Terrain Transformer

Hang Lai, Weinan Zhang, Xialin He, Chen Yu, Zheng Tian, Yong Yu, Jun Wang
IEEE International Conference on Robotics and Automation (ICRA) 2023
(in press)


In this paper, we propose Terrain Transformer (TERT), a high-capacity Transformer model for quadrupedal locomotion control on various terrains. Furthermore, to better leverage Transformer in sim-to-real scenarios, we present a novel two-stage training framework consisting of an offline pretraining stage and an online correction stage, which can naturally integrate Transformer with privileged training. Extensive experiments in simulation demonstrate that TERT outperforms state-of-the-art baselines on different terrains in terms of return, energy consumption and control smoothness. In further real-world validation, TERT successfully traverses nine challenging terrains, including sand pit and stair down, which can not be accomplished by strong baselines.

OmniWheg: An Omnidirectional Wheel-Leg Transformable Robot

Ruixiang Cao, Jun Gu, Chen Yu, and Andre Rosendo
IEEE International Conference on Intelligent Robots and Systems (IROS) 2022 (In press)


We design a novel wheel-leg robot consisting of a separable omni-wheel and 4-bar linkages, allowing the robot to transform between omni-wheeled and legged modes smoothly. In wheeled mode, the robot can move in all directions and efficiently adjust the relative position of its wheels, while it can overcome common obstacles in legged mode, such as stairs and steps. Unlike other articles studying whegs, this implementation with omnidirectional wheels allows the correction of misalignments between right and left wheels before traversing obstacles, which effectively improves the success rate and simplifies the preparation process before the wheel-leg transformation. Our results confirm that this mobile platform can overcome common indoor obstacles and move flexibly on the flat ground with the new transformable wheel-leg mechanism, while keeping a high degree of stability.

Rearranging the Environment to Maximize Energy with a Robotic Circuit Drawer

Xianglong Tan, Zhikang Liu, Chen Yu, and Andre Rosendo
IEEE International Conference on Robotics and Biomimetics (ROBIO) 2022 (In press)


Robots with the ability to actively acquire power from surroundings will be greatly beneficial for long-term autonomy and to survive in uncertain environments. In this work, we present a robot capable of drawing circuits with conductive ink while also rearranging the visual world to receive maximum energy from a power source. A range of circuit drawing tasks is designed to simulate real-world scenarios, including avoiding physical obstacles and regions that would discontinue drawn circuits. We adopt the state-of-the-art Transporter networks for pick-and-place manipulation from visual observation. We conduct experiments in both simulation and real-world settings, and our results show that, with a small number of demonstrations, the robot learns to rearrange the placement of objects (removing obstacles and bridging areas unsuitable for drawing) and to connect a power source with a minimum amount of conductive ink.


May 2022 – Sept. 2022

Transformers on Quadruped Robots

Shanghai Jiao Tong University / University College London

Did an internship at Shanghai Digital Brain Lab. Applied Transformer to quadruped robots.

Feb. 2020 – May 2022

DJI RoboMaster University AI Challenge

ShanghaiTech / Oxford / Cambridge University

Led the team with the universities of Oxford and Cambridge, and update the SLAM modules of the robots. Our team received an A on the technical report, the third prize, and the Academic Incentive Award. Video here.

Mar. 2018 – Sept. 2019

Lip-sync based on Deep Reinforcement Learning

University of Electronic Science and Technology of China

Supervised by Prof. Ning Xie

Combined Actor-Critic and SeqGAN to implement lip animation generation.

June 2019

Temperature-Controllable Smart Clothes

Tsinghua University

Winter camp

Successfully made a prototype with Arduino, Peltier module and temperature sensors and received positive feedback from the entrepreneur community and cooperation invitation from Revsmart Wearable HK Co., Ltd.


Sept. 2020 – Present

ShanghaiTech University

M.Phil. in Computer Science

Expected graduation date: June 2023. Supervised by Prof. Andre Rosendo and Prof. Laurent Kneip.

Mar. 2018 – June 2020

University of Electronic Science and Technology of China

B.Econ. in Finance

Double Degree Programme, where I received my third Bachelor's degree certificate.

Sept. 2016 – June 2020

University of Electronic Science and Technology of China

B.Eng. in Electronic and Information Engineering

Joint School with University of Glasgow.

Sept. 2016 – June 2020

University of Glasgow

B.Eng. in Electronics and Electrical Engineering

First-Class Honour.


Outstanding Student 2022

3%, awarded by ShanghaiTech University.

Outstanding Student 2021

3%, awarded by ShanghaiTech University.

Outstanding Undergraduate Thesis Award 2020

Awarded by University of Electronic Science and Technology of China.

Model Student Scholarship 2018

Awarded by University of Electronic Science and Technology of China.

Outstanding Student Scholarship 2017

Awarded by University of Electronic Science and Technology of China.