Automatic Control & Systems Engineering

Automatic Control & Systems Engineering
Autonomous Systems
Uninhabited air vehicles (UAVs) are increasingly being used in areas where manned aircraft are too dangerous, costly or difficult to deploy.
In addition to well publicised military applications they can be used in many civilian areas such as search and rescue, monitoring of forest
fires, traffic congestion monitoring, aerial surveys, sensing of airborne pollutants, railway track inspection, police and security surveillance.
The development of completely autonomous UAVs presents numerous and formidable research challenges. To solve such a challenging
systems engineering problem requires a truly multi-disciplinary approach. Researchers in the Department of Automatic Control & Systems
Engineering are developing novel intelligent approaches to the design and control of UAVs.
Mini-UAV for Urban Environments
Future uninhabited air vehicles (UAVs) will increasingly be autonomous,
requiring reliable flight control laws for safe operation. This is particularly
challenging in urban environments where the UAV will have to operate in
enclosed spaces. A number of important applications will require this, e.g.
fire and natural disaster search and rescue, police and security services.
The aim of this project is to develop a low-cost mini-UAV (MAV) platform
suitable for operation in urban environments.
A low-cost, fully-autonomous, prototype quadrotor MAV has been
developed. This utilises MEMS gyroscopes and accelerometers to provide
6 degree-of-freedom inertial measurements. An autonomous control
system is implemented on a microcontroller, the output of which controls
the motor speeds providing control in roll, pitch, yaw and vertical
displacement. A full Matlab simulation of the quadrotor suitable for
control-law prototyping has also been developed together with a Matlab
interface for telemetry information.
A second-generation quadrotor MAV is now under development. This will utilise a COTS inertial measurement unit, together with a GPS
receiver, ultrasonic sensors and stereo-cameras as the primary sensors. These will enable a detailed picture of the urban operating
environment to be developed. To provide sufficient processing power the MAV will include dual DSP-FPGA boards for vision processing,
sensor fusion and autonomous control. Live telemetry and video will be transmitted over a wireless LAN to a local base station. Four highpowered brushless motors will provide the necessary lift for the approximate 1.5kg mass. The power for the MAV will be provided onboard
using lithium polymer batteries.
Once the MAV is complete it will provide a test-bed for the development of novel autonomous control algorithms. In particular new sensor
fusion algorithms will be developed which combine the disparate sensor types optimally to provide the MAV with situational awareness and
operational capability in different urban environments including indoors.
Autonomous Control of Multi-UAV Platforms
UAV missions will increasingly rely on the use of multiple platforms,
requiring autonomous navigation, complex mission planning, task
allocation and co-ordination of flight operations. This is particularly true of
search and rescue and aerial surveillance applications. The aim of this
project is to develop novel algorithms for the simulation and control of
multi-UAV platforms.
Initial work has focused on adapting an existing aircraft simulation model
(ADMIRE) for completely autonomous control.
A synthetic flight
management tool has been developed which allows flight plans to be input
and autonomous control strategies automatically generated. These are
based on way-point navigation and use detailed models of aircraft
dynamics. Current work is developing this framework to allow for
automatic collision avoidance and the simulation of multiple-UAVs over a
network of PCs. Future work will focus on developing an intelligent agentbased approach to the autonomous control of multi-UAV platforms. This
will take account of wireless communications, models of which will be
included in the simulation environment.
The algorithms developed in this project will be tested in simulation and also using a group of model helicopters equipped with miniature
autopilots developed in house. A fundamental aim of the project is to develop approaches to the autonomous control of multi-UAV
platforms which are completely scalable and can therefore be applied to various sizes of UAV platforms.
For more information contact:
Dr Tony Dodd
Department of Automatic Control & Systems Engineering
The University of Sheffield
Sheffield, UK