Recent advances in sensor devices, communications, and battery
technology have made fixed-wing Unmanned Aerial Vehicles (UAVs)
smaller in size and cost-competitive. Small-size UAVs are becoming an
increasingly attractive solution for a variety of scientific, civil,
and military applications. While some autonomous UAVs are employed
successfully in security and military services, urban applications
such as infrastructure monitoring demands small or even micro UAVs to
maneuver within complex obstacle-filled environments.
Operating a UAV under these conditions poses a number of difficult challenges.
Environments cluttered with buildings and overhangs require high
maneuverability and fast adaptation to dynamic and unknown obstacles.
Fixed-wing UAVs require a relatively high minimum forward velocity to
maintain lift. Thus, in order to respond quickly to unknown
obstacles, high-performance real-time perception, motion planning, and control
that respect the complex dynamic constraints of the UAV must be accomplished
without any significant delays.
We are focusing on the following reseach topics and goals:
We are currently participating
- Robust real-time computer vision algorithms, especially sensor-fused structure from motion(SFM)
- Efficient motion planning method under dynamic constraints
- A highly realistic simulator (sensors and wireless communications in the loop)
- A Commerical off-the-self(COTS) based system
- Demonstration of with the 3D air slalom scenario
3D Air Slalom Scenario
We have been developing an Unmanned Aerial vehicle System
designed to accomplish a 3D air slalom scenario shown below.
In this scenario, several labeled gates are
arranged in the environment. The gates are placed either on the
ground or in the air, and the UAVs are instructed to pass through each
of the gates sequentially in the order specified.
Many system components comprise the complete solution.
They includes, for examples, gate detection, sky/ground recognition, and
3D environment reconstruction, obstacle-free motion planning,
waypoints controller and so forth.
The UAV maintains up-to-date perception of the environment
and its own state through onboard cameras, IMU and GPS.
Each slalom gate requires the UAV to pass through the target hoop with
the correct 3D position as well as aligned pitch and yaw angles.
Our goal is to build a real-time UAS for 3D slalom
scenarios that allow the UAV to operate reliably in the presence of
fixed or moving obstacles and in even unknown environment.