Embodied Navigation
Closing the loop from perception and localization to decisions and action on real robots.
Robotics · Embodied AI · 3D Scene Intelligence
Yuchen Fan
Building reliable systems from perception to action.
An undergraduate in Robotics Engineering at Beijing Institute of Technology, focused on autonomous driving, robot navigation, 3D scene representations, AI inference deployment, and reliable engineering systems.
01 · Current identity
From research questions to reliable systems
Autonomous Systems
Reliable localization, planning, and systems integration in complex environments.
3D Scene Intelligence
Using 3D representations for navigation, localization, and editable scene generation.
AI Systems Engineering
Turning model capability into deployable, debuggable, and maintainable systems.
02 · Selected systems
Featured projects
Batch-LIO
An engineering and research record for LiDAR–inertial odometry, with explicit system boundaries and evidence.
View projectFormula Student Driverless
Formula Student Driverless experience in the context of autonomous-system integration and track engineering.
View projectUnitree Robotics Systems
Robotics systems records spanning G1 target following and Go2 VLA/VLN deployment.
View project3D Scene Intelligence
A high-level view of 3D representations for instance navigation, visual localization, and editable road scenes.
View project03 · Research journey
A trajectory, not a list of labels
- 2024
Robotics foundations, perception, and control
Completed - 2025
RoboMaster navigation and Formula Student Driverless
Completed - 2026
Embodied navigation, 3DGS, world models, and AI systems
Active
AI InfrastructureRobotics MiddlewareDeveloper Tools
- vLLM
- OpenRLHF
- NVIDIA CCCL
- CycloneDDS
- ROS 2
- OpenCV
Contact
Build, test, and learn in the open.
For research exchange, engineering collaboration, or open-source work.