Modified UAV with actuators enabling rotor tilt.
Investigators at the Institute of Systems Optimization (ITE) at the Karlsruhe Institute of Technology (KIT) in Germany have undertaken a project to increase the level of control of quadrotor drones by adding tilt-able rotors and associated control systems.
The object is to maintain a certain orientation of the UAV and its payload without altering platform attitude, to manage maneuvering more effectively and to compensate for disturbances faster and possibly enlarge the area of operation for rescue forces.
For fire disaster recovery, hovering multi-rotor UAVs can provide invaluable information within buildings, rather than risking the lives of first responders. Locating survivors or difficult to find fire sources using video transmitted by drones may save time and reduce exposure for critical personnel.
A two-part nonlinear control system has been implemented by ITE — the first part takes the measurements of the vehicle dynamics and connects these measurements to a back-stepping controller to generate the desired forces and torque to change vehicle motion.
At first the commanded signals have to be fed through a filter in order to provide smooth and continuous command signals and to produce the derivatives required by the control algorithm. The smoothed command signal is then used by an arbitrary controller to create vectors of required forces and torque to control the attitude and velocity of the vehicle.
Desired force and torque is fed into an adaptive and dynamic control allocation algorithm to generate the values for the actuators – there are four propulsion motor commands and four servo motor commands. The control allocation algorithm is an adaptive algorithm – used in order to adjust for changing situations and environments. For example, when flying in a hallway and near walls, ceiling or floor, flight characteristics change significantly due to different aerodynamic effects. On the other hand, outdoors flight behavior is usually much easier to manage as the only nonlinear behavior occurs relatively close to the ground.
- 3D modeled performance versus flight data.
- 3D modeled performance versus flight data (both diagrams show the same flight).
In order to verify the performance of the system it was modeled — flight dynamics and operator control inputs were simulated. Performance was found to closely match actual recorded flight data. This novel approach could have a number of possible applications — possibly to serve as an alternative to a gimbal mount for a camera?