Supported by the Faculty of Engineering · PolyU

PolyU FENG
AI & Robotics Club

Build. Learn. Compete. Create impact.

A student engineering community that builds intelligent drones, robot dogs and unmanned vehicles with embodied AI, and enters international robotics competitions, guided by PolyU staff, the Industrial Centre, an External Advisory Board and PhD-level mentors.

Join Us Our Projects Q&A

An industrial inspection drone with a zoom and thermal camera — Drones

A quadruped robot walking on the lawn at PolyU — Robot dogs

Autonomous test cars on the PolyU campus — Unmanned vehicles

4Technical groups · mechanical, electronics, perception, software & AI
3Member tiers with public promotion criteria
203Applications in the 2026 recruitment
10+Staff-level supervisors and PhD-level mentors
2027Robocon target · Hong Kong contest to ABU Robocon

Our Vision

About the club

To build a world-class robotics club, first in Hong Kong, then in Asia and the world, and to cultivate outstanding students who can take a real problem all the way to a working, reliable robot. Through our efforts in robotics, we want to make a difference, with trustworthy robotic systems that people can not only use, but also trust.

A robot that works once in a demo is not enough. A trustworthy robot is safe, robust when conditions change, and honest about what it can and cannot do, so that people can rely on it in daily life. This idea is shared with the Trustworthy AI and Autonomous Systems Laboratory (TAS Lab) at PolyU, whose researchers mentor our members.

The club is demanding by design. We train and work to the standard of leading robotics teams at MIT and Stanford, and members who commit gain world-class technology, mentorship, industry connections and a route to PhD study. What you gain

What We Work On

All projects

We build intelligent robots, not only machines. Every platform runs the same embodied AI stack, so what members learn on a drone carries over to a robot dog or a self-driving vehicle.

The common core

Embodied AI: robots that see, think and act

Embodied AI puts artificial intelligence into a physical body. The robot senses the real world, builds an understanding of it, decides what to do and acts, then learns from the result. Models are trained in simulation and tested on real hardware.

PerceiveCameras, LiDAR, GNSS and IMU; deep-learning detection and segmentation
UnderstandSLAM, 3D maps and scene understanding
DecidePath planning, reinforcement learning and vision-language-action models
ActReal-time control on real robots, from simulation to reality

🚁 Drones

Autonomous flight for facade inspection and infrastructure patrol in a high-rise city, with AI defect detection and navigation close to buildings where GNSS is unreliable.

Defect detection
Swarms
Urban navigation

🐕 Robot dogs

Legged robots that learn to walk with reinforcement learning in simulation, then guide visually impaired people, climb stairs and inspect sites. Members also build their own robot dog.

RL locomotion
Sim-to-real
Guide dog

🚗 Unmanned vehicles

Self-driving ground robots for campus delivery, patrol and mapping, using LiDAR and camera perception, learning-based planning and control.

LiDAR SLAM
End-to-end driving
Multi-robot

🏆 Competitions

Robocon 2027: a manual robot and a fully autonomous robot that must find, pick and place blocks on its own. The route runs from the Hong Kong contest to ABU Robocon in Indonesia.