CV

Ziqi Ma

• Email: ziqi_ma0605@163.com
• Phone: +33 06 27 57 46 33
• Address： 828 Bd des Maréchaux, 91120 Palaiseau, France

AREAS OF INTERESTS

• Reinforcement Learning, Imitation Learning, Humain Robot Interaction, Long-horizon Robot Task

EDUCATIONS

• Nov. 2024 - Present    Doctorate Candidate in U2IS, ENSTA(National School of Advanced Techniques), Institut Polytechinique de Paris, Paris, France
  • Research domain: Imitation Learning, Reinforcement Learning, Robotics

• Sep. 2017 - Mar. 2024    Engineer Diplome, SJTU-Polytechnic, ENSTA Paris, Telecom Paris, Mines Paris, Sino-French cooperation

• Sep. 2021 - Mar. 2024    Master Degree, Paris Elite Institute of Technology, Shanghai Jiao Tong University, Shanghai, China
  • Major: Electronic Engineering
• Aug. 2021 - Sep. 2022    Exchange in ENSTA-Paris, Institut Polytechnic of Paris, Paris, France
  • Major: Information and Communication Sciences and Technologies
• Sep. 2017 - Jun. 2021    Bachelor Degree, Paris Elite Institute of Technology, Shanghai Jiao Tong University, Shanghai, China
  • Major: French & Information Engineering

PUBLICATIONS

• Ma, Z., Nguyen, S. M., and Xu P. Leveraging LLM with Active Imitation Learning of Hierarchical Reinforcement Learning Using Emergent Symbolic Representation. Submitted to: 2026 IEEE International conference on robotics and automation (ICRA).

• Ma, Z., Nguyen, S. M., and Xu P. (2025). Bridging Symbols from Language and Hierarchical Reinforcement Learning with Active Imitation. NeurIPS 2025 Workshop on Bridging Language, Agent, and World Models for Reasoning and Planning.

• Ma, Z., Nguyen, S. M., and Xu P. (2025). Can LLMs Translate Human Instructions into a Reinforcement Learning Agent’s Internal Emergent Symbolic Representation? RO-MAN 2025 1st Workshop on Interactive Task Learning in Human-Robot co-construction (ITL4HRI).

• Ma Z., Tian C, Gao Y.(2025) Manipulate as human: learning task-oriented manipulation skills by adversarial motion priors., Robotica, pp. 1–13. doi:10.1017/S0263574725001444.

• Annabi, L., Ma, Z., and Nguyen, S. M. (2024). Unsupervised Motion Retargeting for Human-Robot Imitation. Companion of the 2024 ACM/IEEE International Conference on Human-Robot Interaction (587–591). Association for Computing Machinery (rank A conference)

INTERSHIPS

• Jul. 2023 - Nov. 2023    ABB Mega Factory, Shanghai, China | Department: CRR, R&D
  • Sign language learning by dual-arm robot.
  • Generating sign language movement by “Word-to-Sign” application.
  • Using deep neural network to transfer the movement of human to robot YuMi.
• Jun. 2020 - Sep. 2020    Air Liquide Company, Shanghai, China | Department: Data Science, R&D
  • Optimizing distribution plan.
  • Finding the optimal parameters tuning of clustering algorithm, DBSCAN, to gather the client with similar characteristic.

RESEARCH EXPERIENCES

• Nov. 2024 - Present    Active Learning and Hierarchical Reinforcement Learning | Supervisor: Sao Mai NGUYEN and Philippe XU | U2IS, ENSTA, IP-Paris
  • Research on a Spatial and Temporal HRL algorithm.
  • Research on the alignment capacity of LLM on human language and emergent learning agent.
  • Combining LLM and Active Imitation Learning into HRL to improve training performance.
• Oct. 2023 - Mar. 2024    3D manipulation field descriptor | Supervisor: Yue GAO | AI Institute, Shanghai Jiao Tong University
  • Using cameras to construct a 3D descriptor field for complex manipulation scenario.
  • The 3D field descriptor can output semantic and geometric information, including instance masks, dense semantic features and signed distance to the surface, related to arbitrary point in real 3D world coordinate
  • Tracking the dynamic trajectory of moving object based on semantic and geometric information
• Oct. 2022 - Sep. 2023    Human-like manipulation skill learning | Supervisor: Yue GAO | AI Institute, Shanghai Jiao Tong University
  • Detecting the keypoint of human joint and tools on human skill video clips as motion prior.
  • Learning human-like manipulation policy by RL algorithm with adversarial motion prior.
  • The design of reward function is the sum of goal reward and style reward, where goal reward represents whether the action achieves the goal and style rewards represent whether the state transition has the same distribution of motion priors.
• May. 2022 - Aug. 2022    Unsupervised Human-robot motion retargeting | Supervisor: Sao Mai NGUYEN | U2IS, ENSTA, IP-Paris
  • Assembling robot Poppy.
  • Detecting human skeletons of collected video by Blazepose library.
  • Comparing the results detected by Blazepose with those by Kinect, Openpose, Vicon.
  • Constructing self-supervised network to transfer movement between two personages.
• Jan. 2021 - Jun. 2021    Research on two person emotion interaction detection and generation on video sequences | Independent Research | Shanghai Jiao Tong University
  • Using LSTM network and human interaction dataset to predict the next movement of one agent.
• Feb. 2020 - Jun. 2020    Research about 3D point-cloud reconstruction of SJTU campus virtual map | Independent Research | Shanghai Jiao Tong University
  • Studying the pose of camera, using SIFT and ORB algorithms to find and match feature points of sequential frame

WORK EXPERIENCES

• Sep. 2025 - Present    Teaching Assistant | Machine Learning | ENSTA, IP-Paris
  • Main work: Responsible for one group of tutorials (TD), correcting the exercise and answering questions.
• Mar. 2025 - Jun. 2025    Teaching Assistant | C Programming Language | 21h | ENSTA, IP-Paris
  • Main work: Proposing C Project, answering questions, evaluating and noting.
• Mar. 2023 - Jun. 2023    Teaching Assistant | Mobile Robotics | 15h | SPEIT, Shanghai Jiao Tong University
  • Main work: Teaching students exercise courses, helping teachers correct homework, proctoring the exam.

SUMMER SCHOOL EXPERIENCES

• Sep. 1 - Sep. 5, 2025    PRAIRIE AI Summer School (P.A.I.S.S.) | Grenoble | France
  • Lectures of 19 hours and tutorial of 4 hours, Comprehensive learning on Robotics Control and Optimization; Retrieving, Generating, and Refining for Recognition; Diffusion Flows in Machine Learning; Reinforcement Learning; AI Security; AI Ethics; Data-Driven 3D Vision; Video Understanding Out of the Frame; Weakly Supervised Multi-Label Prediction.
• Jun. 22 - Jun. 28, 2025    Treatment of Signal and Images | Peyresq | France
  • Courses of 21 hours, Comprehensive leaning on Bayesian Estimation and Uncertainty Quantification; Sensibility Analysis, Variance Decomposition; Interpretability of Machine Learning; Conformal Prediction; Gaussian Process; Sequential Learning.
• Jun. 30 - Jul. 5, 2024    ETH RobotX Summer School | Epeisses | Switzerland
  • Lectures of 8 hours, tutorials of 10 hours, research talks of 3 hours and student hands-on work on robotic platforms of at least 28 hours, with a focus on state estimation, localization and mapping, trajectory optimization, path planning, exploration and camera modeling and geometry.
• Jun. 8 - Jun. 10, 2024    AERO-TRAIN Summer School | Chania | Greece
  • Lectures on Perception, planning, control and Human-Robot Interaction.

HONORS & AWARDS

• Jun. 2021    Excellent Bachelor of Shanghai Jiao Tong University
• Oct. 2020    Excellent Scholarship of SPEIT, Shanghai Jiao Tong University (5%)
• Oct. 2020    Certificate of Excellence in Operational Internship
• Dec. 2019    A Prize of Shanghai Jiao Tong University (10%)
• Apr. 2019    Honorable Mention in 2019 MCM
• Dec. 2018    B Prize of Shanghai Jiao Tong University (15%)
• Oct. 2018    Three Good Student of Shanghai Jiao Tong University

SKILLS

• Language: French, English, Chinese
• Programming: C/C++, Python, Java
• ML systems: Pytorch
• Physical Simulation Platforms: Mujoco, Isaac Gym, Isaac Sim
• Data Acquisition & Processing: Matlab