Slides PDF

Zongyao Jin

Roboticist · Ph.D.

zongyaojin@outlook.com zongyaojin zjin.pages.dev

Experience

Beijing Science & Technology Project Manager Management Co. Ltd., Beijing, China
2025 - Present
  • Phoenix Project Manager
    • Supporting the establishment and development of key research institutions such as national labs and new R&D entities, as well as key regions such as the Three Cities and One District, etc.
    • Phoenix Project was initiated by Beijing Municipality Government, administered by Beijing Municipal Science & Technology Commission, with Beijing Science & Technology Project Manager Management Co. Ltd. serving as its implementing platform.
    • Government announcement: https://www.beijing.gov.cn/zhengce/zwmc/202201/t20220124_2598556.html
ByteDance, Seed Department, Beijing, China
2024 - 2025
  • Staff Robotics Scientist
    • Developed and maintained reactive whole-body control, optimal control, model predictive control, impedance/admittance control, trajectory generation, and software infrastructure for humanoids with VLA and tele-operation, etc.
    • Received 2 excellent team performance awards as a member of different teams.
Flexiv Robotics, Santa Clara, CA, US
2020 - 2024
  • Senior Robotics Controls Engineer, 2022 - 2024
  • Robotics Controls Engineer, 2020 - 2022
    • Developed whole body controllers, impedance, motion-force, visual-servo control tasks for flexible joint robots with redundant DOF.
    • Developed robotic software architecture for various robots to perform manipulation tasks in unstructured environments.
    • Acquired 1 patent, implemented and delivered 5 robotic software/algorithm products, received Beyond Expectations in annual performance review for 2 consecutive years.
Texas A&M University, College Station, TX, US
2015 - 2020
  • Ph.D. Student
    • Ph.D. research funded by NSF grant 1527828; proposed task learning, intent prediction, and blended shared control methods that allow robots to learn from demonstration and collaborate with humans.
    • Published 9 peer-reviewed academic papers as the first author, 2 in IEEE Transactions on Control Systems Technology; Ph.D. thesis included as an independent chapter in the book of Recent Advances in Industrial Robotics; received 2 highest-level student awards from the graduate program.
China Academy of Space Technology, Beijing, China
2012 - 2015
  • Aerospace System Engineer
    • Participated in the research and development of various space robotic systems.
    • Published 2 peer-reviewed academic papers, acquired 3 patents.

Education

Texas A&M University, College Station, TX, US
2015 - 2020
  • Ph.D. in Mechanical Engineering
  • Thesis: Novel Methods for Human-Robot Shared Control in Collaborative Robotics.
China Academy of Space Technology, Beijing, China
2009 - 2012
  • M.S. in Aerospace Engineering
  • Thesis: Dynamics and Path Planning for Large Redundant Space Manipulators with Link Flexibility.
Harbin Engineering University, Harbin, Heilongjiang, China
2005 - 2009
  • B.S. in Mechanical Engineering

Honors

2025 - National High-Level Young Talent, National Ministries and Commissions, China.
2024 - ByteDance Semi-Annual Team Excellence, ByteDance, China.
2024 - ByteDance Quarterly Team Excellence, ByteDance, China.
2022 - Flexiv Robotics Annual Performance Excellence, Flexiv Robotics, USA.
2021 - Flexiv Robotics Annual Performance Excellence, Flexiv Robotics, USA.
2018 - Jame J. Cain Outstanding Graduate Student Achievement (0.2% award rate), Texas A&M University, USA.
2017 - Jame J. Cain Graduate Teaching Excellence (0.2% award rate), Texas A&M University, USA.
2009 - Provincial Outstanding Undergraduate Student (0.2% award rate), Heilongjiang Province Ministry of Education, China.
2008 - National Undergraduate Scholarship (0.2% award rate), National Ministry of Education, China.

Skills

Analytical: Software Architecture, Robot Manipulation, Dynamics, Kinematics, Whole-Body Control, Visual-Servo Control, Hybrid Motion-Force Control, Impedance/Admittance Control, Optimal Control, State Estimation, Online Trajectory Generation

Languages: C++, CMake, Python, Bash, HTML, CSS, JavaScript, Liquid

Frameworks: ROS-2, Eigen, qpOASES, RBDL, RBDyn, Bullet, Gazebo, Chai3d, glog, Poetry, UV, FastDDS, gRPC, Qt, Three, Jekyll, npm, Vite

Projects

Hybrid Motion-Force-Impedance Control for Generic Peg-In-Hole Manipulations
@ Flexiv Robotics
  • Developed manipulation strategies for robots to reliably perform 0.1mm tolerance peg-in-hole under small uncertainties.
  • Developed high performance operational space motion-force control algorithms that handle (1) dynamics model uncertainties, (2) robot singularity avoidance, (3) flexible joint dynamics, (4) impedance adjustment, (5) in-task nullspace decoupling projection.
  • Developed online-generated self-adjusting trajectories that ensure 100% area coverage when performing peg-in-hole search.
  • The developed package became one of Flexiv's highlighted technologies and is being used in all assembly applications.
  • Demo: https://zjin.pages.dev/tech/demo/2023/10/21/peg-in-hole.html
Image-Based Visual Servo Control for High Precision Flexible Printed Circuit Assembly
@ Flexiv Robotics
  • Developed manipulation strategies for robots to reliably perform 0.01mm tolerance assembly under large uncertainties.
  • Developed high performance image-based visual servo control algorithms that integrate (1) image Jacobian singularity avoidance, (2) feature depth estimation, (3) optimal feature sets alignment, (4) image space boundary protection, (5) Cartesian impedance blending.
  • Developed a complete vision-based robot assembly solution by coordinating a team of AI and mechanical engineers to improve image detection quality and gripper design around the vision control algorithm limitations.
  • The developed package became one of Flexiv's highlighted technologies and is being used in all FPC assembly applications.
  • Demo: https://zjin.pages.dev/tech/demo/2023/10/08/ibvs-control.html
Control Software Framework Redesign for Heterogeneous Robots and Motor Control Modes
@ Flexiv Robotics
  • Restructured the company's control software framework to support whole body control of (1) parallel-link robots, (2) serial-link robots, (3) robots with linear rails or mobile platforms, (4) macro-mini robots, (5) dual-arm robots, (6) robot with mixed position and torque controlled joints, from 3-DOF to 15-DOF.
  • Redesigned the company's (1) robot description data structure to handle robots with modular joint groups via URDF and SRDF, and (2) robot behavior system to allow users to dynamically switch task priorities and coordinate different joint groups.
  • Refactored the company's control software architecture, optimized composition, inheritance, interface, factory structures to manage multiple dynamics, whole body controllers, control tasks, and robot behavior primitives with unified interfaces and minimal dependencies.
  • Migrated the company's all existing controller modules to work with the new framework, maximized software extendibility.
  • Demo: https://zjin.pages.dev/tech/demo/2023/08/07/control-framework.html
Physics-Based Simulator in C++ for Robotics Research and Development
@ Flexiv Robotics
  • Designed and implemented a robot simulator in C++ with GUI, 3D graphics, data plots, dynamics, contact and collision physics.
  • Designed the simulator's software architecture using various design patterns such that the dynamics, controllers, physics, graphics are polymorphic, decoupled, and scalable.
  • Implemented whole body control, image-based visual servo control, position-based visual servo control, operational space motion-force control, impedance-admittance control, and online trajectory generators for Rn and SE(3).
  • Verified various robot motion and control algorithms in the simulator before they are deployed to real Flexiv robots.
  • Demo: https://zjin.pages.dev/tech/demo/2023/05/21/robot-simulator.html
Robot Behavior Generation Software Toolbox for Semi-Repetitive Manipulations
@ Flexiv Robotics
  • Analyzed common patterns of semi-repetitive tasks and extracted the key components for building a generic software toolbox.
  • Designed a software toolbox for clients to program semi-repetitive robot behaviors, implemented the robot behavior generation logic.
  • Led the toolbox development with a team consists of UX designers, UI, software, and control engineers.
  • The developed toolbox became one of Flexiv's "in-app purchase" software modules and is being used in all palletization-like applications.
  • Demo: https://zjin.pages.dev/tech/demo/2023/07/16/palletizing.html
Novel Methods for Human-Robot Shared Control in Collaborative Robotics
@ Texas A&M University
  • Proposed a set of theories to make construction robots more intelligent and collaborative; analogous to semi-autonomous driving, but construction machines work in more complicated environment and are more difficult to operate, these methods can help human operators control these machines without significant skills and training, and provide online intelligent assistance.
  • Proposed and implemented methods for task learning from demonstration, operator intent prediction, control autonomy sharing and arbitration, online task model adjustments.
  • Demo: https://zjin.pages.dev/tech/demo/2020/08/01/phd-research.html
Personal Website with Jekyll, Python, Ruby, Liquid, HTML, CSS, Javascript
@ Extracurricular
  • Designed and implemented a personal website to organize study notes and fragments of life, and enjoy coding something just for fun.
  • Incorporated Agile Project Management and documentation features inspired by Jira, Confluence, Feishu Doc, blog websites, and academic papers.
  • Based on the concept of Model-View-Controller, developed data, templates, control logic modules that are decoupled and scalable.
  • Demo: https://zjin.pages.dev

Publications

[1] Z. Jin, P. Pagilla, "Shared Control with Efficient Subgoal Identification and Adjustment for Human-Robot Collaborative Tasks", in IEEE Transactions on Control Systems Technology, 2022.
[2] Z. Jin, P. Pagilla, H. Maske, G, Chowdhary, "Task Learning, Intent Prediction, and Adaptive Blended Shared Control with Application to Excavators", in IEEE Transactions on Control Systems Technology, 2021.
[3] Z. Jin, P. Pagilla, "Blended Shared Control in Collaborative Robotics (book chapter)", in Volume 2: Recent Advances in Industrial Robotics, Manufacturing in the Era of 4th Industrial Revolution, 2021.
[4] Z. Jin, P. Pagilla, "Subgoal Learning via Operator Command Quantification for Human-Machine Shared Control Task Modeling", in American Control Conference, 2020.
[5] Z. Jin, P. Pagilla, "Collaborative Operation of Robotic Manipulators with Human Intent Prediction and Shared Control", in IEEE International Conference on Human-Machine Systems, 2020.
[6] Z. Jin, P. Pagilla, "Operator Intent Prediction with Subgoal Transition Probability Learning for Shared Control Applications", in IEEE International Conference on Human-Machine Systems, 2020.
[7] Z. Jin, P. Pagilla, "Human-Robot Teaming with Human Intent Prediction and Shared Control", in Artificial Intelligence and Machine Learning for Multi-Domain Operations Applications, 2020.
[8] Z. Jin, P. Pagilla, H. Maske, G, Chowdhary, "Methods for Blended Shared Control of Hydraulic Excavators with Learning and Prediction", in IEEE Conference on Decision and Control, 2018.
[9] Z. Jin, P. Pagilla, H. Maske, G, Chowdhary, "Blended Shared Control with Subgoal Adjustment", in IEEE International Conference on Systems, Man, and Cybernetics, 2018.
[10] Z. Jin, C. Tan, "A Space Manipulator Path Planning Method with Synthesized Optimization", in Spacecraft Engineering, 2015.
[11] Z. Jin, Y. Liu, C. Tan, "Review and Analysis of the Asteroid Capture and Retrieval Mission", in Spacecraft Engineering, 2014.

Patents

[1] Z. Jin, Y. Chen, A Novel Gripper and Manipulation Strategies for Cable Assembly Operations, Flexiv Robotics, CN 114952915 A, 2022.
[2] A. Zhu, Z. Jin, et al., A Nuclear Powered Manned Mars Spacecraft System, China Academy of Space Technology, CN 206394910 U, 2017.
[3] A. Zhu, Z. Jin, et al., A Manned Mars Spacecraft Exterior Design, China Academy of Space Technology, CN 3034279633 S, 2017.
[4] C. Wei, Z. Jin, et al., A Small Spacecraft Docking System Design, China Academy of Space Technology, CN 103625656 B, 2015.

Hobbies

Piano: Achieved the 8th Grade in Piano Performance at the Central Conservatory of Music of China.

Running: A regular runner, averaging approximately 100 kilometers per month.

金宗耀

博士 · 国家级海外高层次青年人才

zongyaojin@outlook.com zongyaojin zjin.pages.dev

工作经历

北京首都科技项目经理人管理有限公司, 中国北京
2025 - Present
  • 朱雀计划 · 科技项目经理人
    • 主要服务于国家实验室、新型研发机构等重点科研机构和“三城一区”等重点区域的建设与发展。
    • “朱雀计划” 由北京市人民政府发起，北京市科学技术委员会主管，北京首都科技项目经理人管理有限公司为载体执行。
    • 政府文件: https://www.beijing.gov.cn/zhengce/zwmc/202201/t20220124_2598556.html
字节跳动, Seed 部门, 中国北京
2024 - 2025
  • 机器人算法专家
    • 负责开发并维护人形机器人全身控制、最优控制、避障控制、模型预测控制、阻抗/导纳柔顺控制、轨迹生成、VLA模型跟踪、遥操作控制等算法，及相关软件构架与部署。
    • 作为团队/项目组成员，获得 2 项团队/项目组优秀表现奖。
非夕机器人, 美国硅谷
2020 - 2024
  • 高级机器人控制算法工程师, 2022 - 2024
  • 机器人控制算法工程师, 2020 - 2022
    • 负责冗余度机械臂控制算法开发及软件构架与实现，主要包括：全身控制、力/位/视觉混合控制、柔顺控制、精密操作控制。
    • 作为第一作者获得专利 1 项，落地软件/算法产品 5 项，连续 2 年获得“超出预期”年终考核评级。
美国德克萨斯农工大学, 美国德克萨斯州
2015 - 2020
  • 博士研究生
    • 负责美国国家自然科学基金项目（NSF 1527828）“协作机器人人机交互与智能共享控制技术” 的理论及算法研究。
    • 作为第一作者发表论文 9 篇，其中 2 篇为 SCI 一区（中科院二区）期刊论文；博士论文被纳入学术专著成为独立章节；获得 2 项学院最高级别研究生奖项。
中国空间技术研究院总体部, 中国北京
2012 - 2015
  • 航天系统工程师
    • 参与空间机械臂系统、模块化卫星系统、高轨空间目标抓捕系统、核动力载人火星飞船等项目的预研工作。
    • 作为第一作者发表国内核心期刊论文 2 篇，作为共同作者获得专利 3 项。

教育背景

美国德克萨斯农工大学, 美国德克萨斯州
2015 - 2020
  • 机械工程博士
  • 博士论文：协作机器人智能人机共享控制新方法研究
中国空间技术研究院, 中国北京
2009 - 2012
  • 航天器工程硕士
  • 硕士论文：柔性冗余度大型空间机械臂动力学与路径规划技术研究
哈尔滨工程大学, 中国黑龙江
2005 - 2009
  • 机械工程学士

所获荣誉

2025 - 国家级海外高层次青年人才, 国家部委, 中国.
2024 - 字节跳动半年度优秀团队奖, 字节跳动有限公司, 中国.
2024 - 字节跳动季度优秀团队奖, 字节跳动有限公司, 中国.
2022 - Flexiv Robotics 年度优秀个人贡献, 非夕机器人, 美国.
2021 - Flexiv Robotics 年度优秀个人贡献, 非夕机器人, 美国.
2018 - Jame J. Cain 研究生优秀成果奖 (0.2% 获奖率), 美国德克萨斯农工大学, 美国.
2017 - Jame J. Cain 研究生教学优秀奖 (0.2% 获奖率), 美国德克萨斯农工大学, 美国.
2009 - 黑龙江省普通高校优秀毕业生 (0.2% 获奖率), 黑龙江省教育厅, 中国.
2008 - 国家奖学金 (0.2% 获奖率), 国家教育部, 中国.

专业技能

技术领域: 软件构架, 机器人操作, 动力学, 运动学, 全身控制, 最优控制, 视觉伺服/力位混合/阻抗/导纳控制, 状态估计, 轨迹生成

编程语言: C++, CMake, Python, Bash, HTML, CSS, JavaScript, Liquid

软件框架: ROS-2, Eigen, qpOASES, RBDL, RBDyn, Bullet, Gazebo, Chai3d, glog, Poetry, UV, FastDDS, gRPC, Qt, Three, Jekyll, npm, Vite

项目经历

基于笛卡尔空间力位混合控制的机器人高精度通用组装插拔操作算法
@ 非夕机器人
  • 实现了基于笛卡尔空间力位混合控制和在线自适应搜索轨迹的机器人高精度通用组装插拔算法，能够在传统位置控制机器人无法应对的不确定环境中进行 0.1mm 级高精度组装插拔操作，并达到 99.5% 成功率，成为 Flexiv Robotics 核心技术亮点之一，在各大展会及宣传推广中均作为重点进行展示，在公司所承接的各类组装类产线上实现落地应用。
  • 独立开发了相关控制算法、轨迹算法、优化算法，操作策略。与各团队协同完成算法迭代，软件实现，设计测试实验，完成文档编写，获得发明专利 1 项。将算法融合进 Flexiv Robotics 核心控制算法软件库。
  • 演示（视频需VPN）: https://zjin.pages.dev/tech/demo/2023/10/21/peg-in-hole.html
基于视觉伺服与柔顺控制的超高精度机器人组装操作算法
@ 非夕机器人
  • 实现了多种基于位置及图像特征的视觉伺服算法、融合力觉与视觉混合控制，应用于超高精度机器人组装操作，能够在零部件位置未知的情况下进行 0.01mm 级超高精度组装，并达到 99.5% 成功率。克服了传统机器人与控制算法无法应对的工况。成为 Flexiv Robotics 核心技术亮点之一，在各大展会及宣传推广中均作为重点进行展示，在公司所承接的高精度电子产品组装产线上实现落地应用。
  • 独立开发了相关控制算法，视觉算法，操作策略，完成算法迭代。协同团队根据视觉算法能力开展机器人末端工具设计、AI 图像识别能力提升。完成软件实现，文档编写，设计测试实验。将算法融合进 Flexiv Robotics 核心控制算法软件库。
  • 演示（视频需VPN）: https://zjin.pages.dev/tech/demo/2023/10/08/ibvs-control.html
多模态混合类型机器人全身控制软件构架搭建
@ 非夕机器人
  • 重构了 Flexiv Robotics 公司的控制软件框架，可以将机器人全身控制算法应用在 Flexiv Robotics 推出的全系列机器人产品上，包括串联机器人及并联机器人。实现了基于动力学的全身力矩控制算法在不同类型/属性机器人上的无差异化应用。
  • 独立设计了能够同时支持并联与串联机器人的通用控制软件构架，将 Flexiv Robotics 原有的控制任务及算法无缝移植到重构后的新控制软件构架之上。协同各团队通过实验及测试，完成 Flexiv Robotics 核心控制软件构架的迁移与扩展。
  • 演示（视频需VPN）: https://zjin.pages.dev/tech/demo/2023/08/07/control-framework.html
基于 C++ 的机器人物理仿真器
@ 非夕机器人
  • 独立设计、开发了一个基于 C++ 的机器人物理仿真器，包含图形用户界面、三维可视化、数据绘图、动力学、物理碰撞等；设计了仿真器的软件构架，使得动力学、控制器、物理、图形等软件模块耦合度低、易于扩展。
  • 实现了全身控制、基于图像的视觉伺服控制、基于位置的视觉伺服控制、操作空间力位混合控制、阻抗自适应控制、在线轨迹生成等算法；对多种运动控制算法在实机部署前进行了仿真验证，保证了算法的正确性及可靠性，大幅提高了开发效率。
  • 演示（视频需VPN）: https://zjin.pages.dev/tech/demo/2023/05/21/robot-simulator.html
机器人不确定环境下高精度 3C 元器件上下料控制及操作软件
@ 非夕机器人
  • 完成了融合 AI 图像识别、视觉伺服控制、力位混合控制的 3C 产业生产线零件上下料综合操作控制软件工具包。被纳入 Flexiv Robotics 核心软件工具，在公司承担的 3C 产线上广泛使用，在传统机器人靠位置精度难以完成任务的不确定工况下，取代人工进行自动化装配。
  • 独立开发了软件的核心框架，协同团队融合 AI 图像识别与力位混合控制算法，编制操作软件，形成 Flexiv Robotics 的单独付费软件包，被终端客户广泛使用。
  • 演示（视频需VPN）: https://zjin.pages.dev/tech/demo/2023/07/16/palletizing.html
协作机器人的智能人机共享控制新方法研究
@ 美国德克萨斯农工大学
  • 提出了一系列人机协作理论方法，用于创造更加智能/协作的工程类机器人（如挖掘机）；类似车辆辅助驾驶，工程类机器人工作环境及操作方法均存在更高的复杂度，该系列方法可以通过提供在线智能辅助的方式，使不具备大量经验和技能的操作员更容易操作该类机器人。
  • 提出、实现、并通过实验验证了下列理论方法：从演示学习任务模型，操作员意图预测，人机控制权限决策与共享，任务模型在线调整。
  • 演示（视频需VPN）: https://zjin.pages.dev/tech/demo/2020/08/01/phd-research.html
基于 Jekyll, Python, Ruby, Liquid, HTML, CSS, Javascript 的个人网站
@ 个人项目
  • 独立设计、开发、并编程实现了一个个人网站，用于整理学习笔记、记录生活、以及对机器人领域以外的编程技巧和范式进行探索。
  • 参考 Jira/Confluence 敏捷项目管理，飞书文档，博客网站，学术论文等各类信息管理与呈现载体，将可取功能/特性融入个人网站的设计，实现数据的模块化储存及多维呈现。
  • 基于 Model-View-Controller 的软件概念设计了数据结构、内容模板、逻辑单元等，使网站软件模块间逻辑独立、易于维护与扩展。
  • 演示（视频需VPN）: https://zjin.pages.dev

发表论文

[1] 中科院二区（SCI 一区）| Z. Jin, P. Pagilla, "Shared Control with Efficient Subgoal Identification and Adjustment for Human-Robot Collaborative Tasks", in IEEE Transactions on Control Systems Technology, 2022.
[2] 中科院二区（SCI 一区） | Z. Jin, P. Pagilla, H. Maske, G, Chowdhary, "Task Learning, Intent Prediction, and Adaptive Blended Shared Control with Application to Excavators", in IEEE Transactions on Control Systems Technology, 2021.
[3] 学术专著（独立章节）| Z. Jin, P. Pagilla, "Blended Shared Control in Collaborative Robotics (book chapter)", in Volume 2: Recent Advances in Industrial Robotics, Manufacturing in the Era of 4th Industrial Revolution, 2021.
[4] EI 索引 | Z. Jin, P. Pagilla, "Subgoal Learning via Operator Command Quantification for Human-Machine Shared Control Task Modeling", in American Control Conference, 2020.
[5] EI 索引 | Z. Jin, P. Pagilla, "Collaborative Operation of Robotic Manipulators with Human Intent Prediction and Shared Control", in IEEE International Conference on Human-Machine Systems, 2020.
[6] EI 索引 | Z. Jin, P. Pagilla, "Operator Intent Prediction with Subgoal Transition Probability Learning for Shared Control Applications", in IEEE International Conference on Human-Machine Systems, 2020.
[7] EI 索引 | Z. Jin, P. Pagilla, "Human-Robot Teaming with Human Intent Prediction and Shared Control", in Artificial Intelligence and Machine Learning for Multi-Domain Operations Applications, 2020.
[8] EI 索引 | Z. Jin, P. Pagilla, H. Maske, G, Chowdhary, "Methods for Blended Shared Control of Hydraulic Excavators with Learning and Prediction", in IEEE Conference on Decision and Control, 2018.
[9] EI 索引 | Z. Jin, P. Pagilla, H. Maske, G, Chowdhary, "Blended Shared Control with Subgoal Adjustment", in IEEE International Conference on Systems, Man, and Cybernetics, 2018.
[10] 国内核心 | Z. Jin, C. Tan, "A Space Manipulator Path Planning Method with Synthesized Optimization", in Spacecraft Engineering, 2015.
[11] 国内核心 | Z. Jin, Y. Liu, C. Tan, "Review and Analysis of the Asteroid Capture and Retrieval Mission", in Spacecraft Engineering, 2014.

所获专利

[1] 金宗耀（1/2），陈郁聪, 机械手、机械手组件、机器人及装配带线束的插头的方法, 非夕机器人, CN 114952915 A, 2022.
[2] 朱安文，金宗耀（2/12）等, 一种基于兆瓦级核反应堆电源的载人火星飞船, 中国空间技术研究院总体部, CN 206394910 U, 2017.
[3] 朱安文，金宗耀（2/12）等, 载人火星飞船, 中国空间技术研究院总体部, CN 3034279633 S, 2017.
[4] 魏承等，金宗耀（9/9）, 一种小型航天器对接机构, 中国空间技术研究院总体部, CN 103625656 B, 2015.

业余爱好

钢琴: 中国中央音乐学院钢琴演奏八级

跑步: 常年坚持跑步，平均每月跑步约100公里