🎓 M.S. in Control Science & Engineering @ SJTU
🤖 Agent / Embodied AI / Robotics Learning
- Focus on Code Agent and Agent systems for real-world robotics
- Experienced in connecting LLMs → perception → planning → execution
- Building long-horizon task frameworks with structured memory, tool use, and failure recovery
- Background in Reinforcement Learning, Sim2Real, and multi-robot systems
- 🧩 Design Agent runtime systems for robotic tasks
- 🔗 Integrate MCP tools, VLA policy, and execution pipelines
- 🧠 Develop long-horizon planning, task decomposition, and recovery mechanisms
- 🤖 Deploy RL policies from simulation to real-world robotic systems
- Long-horizon robotic task execution with Agent architecture
- Structured memory + tool calling + task state tracking
- LLM reasoning aligned with executable robot actions
🔗 Project: https://github.com/RoboClaw-Robotics/RoboClaw
📄 Paper: https://arxiv.org/abs/2603.11558
- 20-agent real-world cooperative control system
- Vision-based tracking (BoxMOT) + MARL
- Closed-loop control at 30Hz
- RL-based control for magnetic soft robots
- Unity + ML-Agents + Gymnasium pipeline
- Real-world deployment with 0.78mm precision error
- GitHub: https://github.com/zyl-hub
- Email: zhouyunlang@sjtu.edu.cn
Interested in building next-generation Code Agent and Agent-driven robotic systems