Master's student in Automation at South China University of Technology, focused on mechanical manipulation, robot control, reinforcement learning control, and robot learning.
Current project focus:
- Mechanical manipulation and model-based robot control
- Residual reinforcement learning for trajectory tracking
- ROS2 manipulation systems with SO-101 real-robot validation
- MuJoCo / MuJoCo Warp robot-learning experiments
- robust-manipulator-residual-rl - model-based control + bounded Residual PPO under inertial model mismatch.
- so101-visual-tactile-grasp - ROS2 perception, coordinate transform, DLS IK/planning, FlexiTac contact feedback, and real SO-101 grasp-and-lift validation.
- moveit2-panda-learning-demo - MoveIt 2 skill-validation demo with MoveGroupInterface, PlanningScene collision objects, and OMPL obstacle avoidance.
- so101-nexus-reproduction - SO101-Nexus BC + PPO reproduction and cross-backend evaluation notes.
GitHub: adsade5
华南理工大学自动化硕士研究生，方向聚焦机械臂操作、机器人控制、强化学习控制与 Robot Learning。
当前项目重点：
- 机械臂操作与基于模型的机器人控制
- 用于轨迹跟踪的 Residual Reinforcement Learning
- ROS2 机械臂系统与 SO-101 真机验证
- MuJoCo / MuJoCo Warp Robot Learning 实验
- robust-manipulator-residual-rl - 基于模型控制 + 惯性模型失配下的限幅 Residual PPO。
- so101-visual-tactile-grasp - ROS2 感知、坐标变换、DLS IK/规划、FlexiTac 触觉反馈与真实 SO-101 抓取-抬升验证。
- moveit2-panda-learning-demo - MoveIt 2 技能验证 Demo，包含 MoveGroupInterface、PlanningScene collision objects 和 OMPL obstacle avoidance。
- so101-nexus-reproduction - SO101-Nexus BC + PPO 复现与跨后端评估记录。