I'm Lain-Ego, a robotics motion-control engineer and open-source community builder.
I currently work as an early-stage motion-control member at Questech, contributing to its humanoid robots. My work covers motion retargeting, real-time whole-body teleoperation, hierarchical reinforcement-learning policies, and robot perception preprocessing. Previously, I led locomotion deployment for an open-source quadruped at HighTorque Robotics, taking policies from simulation to real hardware.
My core interest is making motion policies run reliably on real robots. I work across mechanical design, embedded systems, simulation, reinforcement learning, and full-stack deployment for humanoids, quadrupeds, wheeled-legged robots, and exoskeletons.
中文版个人简介（点击展开）
你好，我是 Lain-Ego，机器人运动控制工程师与开源社区建设者。
目前作为迁星智能早期成员参与人形机器人研发，主要负责动作重定向、低延迟全身遥操作、强化学习分层运控策略及机器人视觉预处理。此前曾在广州高擎动力主导开源小型四足机器人的运控部署与 Sim2Real 落地。
我关注如何让运动策略在真实机器人上可靠运行，具备机械结构、电控嵌入式、仿真训练、强化学习与实机部署的跨栈研发经验，项目覆盖人形、四足、轮足机器人与主动式外骨骼。
我曾任 ROBOCON 机器人队副队长，累计获得 30 余项国家级、省级竞赛奖项；同时创建 LocoWiki 运控社区并持续维护多个机器人开源项目。
Policies that run. From mechanics and firmware to learning-based control and deployment.
- 🏆 30+ competition awards, including 7 National Second Prizes and 11 National Third Prizes in ROBOCON from 2024 to 2026
- 🦿 Physical AI Hackathon champion in the alloop track with the G-Exo active lower-limb exoskeleton
- 🤖 Motion-control experience across humanoid, quadruped, wheeled-legged, underwater, and exoskeleton robots
- 🌐 Creator of LocoWiki: 100K+ views and a 900+ member robotics community
- 📣 1M+ cross-platform impressions, 2,600+ followers on X, and 180+ GitHub followers
- 🎤 Invited speaker on legged robotics at the 2026 ROBOCON Annual Conference
- Humanoid Motion Control @ Questech — motion retargeting, mocap data pipelines, real-time whole-body teleoperation, hierarchical policies, and URDF-based self-occlusion masks
- S-Dog Wheeled-Legged Robot — a 16-DoF, ~13 kg platform using curriculum learning and reinforcement learning; demonstrated 14 cm obstacle traversal, 40 cm platform climbing, and self-recovery; exhibited at WAIC 2026
- G-Exo Active Lower-Limb Exoskeleton — consumer-oriented 4-DoF active exoskeleton for outdoor hiking; mechanical, electrical, URDF, and learning-control development
-
HighTorque MiniDog — open-source quadruped locomotion stack based on HIMLoco, with sim-to-real deployment, DreamWaQ/AMP reproduction, staged reward strategies, and reproducible documentation.
Reinforcement LearningSim2RealQuadruped -
ROBOCON BRS Parallel Quadruped — 8-DoF parallel quadruped covering mechanical design, kinematics, VMC locomotion, FAST-LIO mapping, and autonomous navigation. Winner of three National Second Prizes at ROBOCON 2025.
ROSC++VMCNavigation -
Bionic Lobster ROV — multi-DoF underwater robot with controllable buoyancy, heterogeneous actuator integration, and vector-thrust omnidirectional control. Provincial Grand Prize; featured by Xinhua News Agency and Guangming Daily.
Embedded SystemsControlUnderwater Robotics -
OpenDoge — crowdfunded open-source quadruped project with a complete simulation-to-real workflow and hands-on learning resources.
Open HardwareQuadrupedCommunity -
LocoWiki — a structured robotics knowledge base covering competitions, papers, open-source projects, simulation, deployment notes, and engineering learning paths.
Knowledge BaseDocumentationCommunity
Reward Design · Curriculum Learning · Domain Randomization · Sim2Real · System Identification · IMU Calibration & Filtering
Building reliable robots at the intersection of mechanics, electronics, control, and AI. 🤖