Karen Feigh (MPhil 2001-2003)
An Airspace Simulator for Air Traffic Management Research
This research describes the design and implementation of a discrete event simulation of air traffic. The model is designed to simulate a large number of aircraft over a wide area much faster than real-time. The aircraft dynamics is taken from the BADA database. Realistic traffic profiles are modelled from published data of departures and the NOAA weather database was used to provide weather inputs. The simulator provides the basis for a conflict resolver. The model also includes a scripting function, which allows the behaviour of air traffic controllers to be specified so that the performance of automated approaches to air traffic management can be compared with human-centred control strategies.
Huamin Jia (PhD 2000-2004)
Data Fusion Methodologies for Multisensor Aircraft Navigation Systems
This PhD was undertaken as part of the EU SHINE project (G4RD-CT-2000-00227) investigating novel inertial system sensor fusion algorithms for redundant navigation systems combing inertial and GPS sensors. The thesis covers the development of equations to represent inertial networks (particular skew-redundant sensors) to formulate distributed data fusion algorithms. A large simulation package was developed in Matlab to provide detailed models for inertial and GPS sensors. An extensive simulation study of a wide range of inertial and GPS sensor accuracies is presented. The research shows improved navigation performance and fault tolerance with low specification sensors.
Anthony Clare (PhD 2002-2006)
Real-time Modelling and Sensor Fusion for a Synthetic Vision System
This PhD was undertaken as part of an EPSRC grant GR/R43020/02 ‘Feature Extraction in Enhanced Vision Systems for Civil Aircraft’, in collaboration with the University of Essex and BAE Systems. A forward-looking radar was developed by BAE Systems which penetrates cloud and rain for 4-8 Km to provide a radar image in a Head-Up Display (HUD) to enable flight crews to proceed with a landing after reaching the 200 feet decision point of an ILS approach. The thesis describes the development of a real-time radar model which can be parameterised to model specific radar receiver properties. The model uses a standard OpenFlight database, modified to include radar properties of the terrain. SGI Performer provides the 3D imaging of the radar display in the HUD which is augmented with real-time software to incorporate the correct radar characteristics in the image. By combining outputs from a radar tracker (used to locate the runway features in a cluttered image) and an inertial navigation system (INS), a Kalman Filter is used to smooth abrupt changes in the tracking loop and eliminate bias in the INS measurements. The performance of the Kalman Filter is monitored to detect errors and to alert pilots to system failure if a threshold is exceeded.
Beini Wu (MPhil 2002-present)
Aircraft Conflict Resolution Methods
In future air traffic environments, where aircraft broadcast their position, speed, heading and intent regularly, using GPS for navigation, the resolution of conflicts may be computed as a distributed network of aircraft. By making small heading or speed changes, and subsequently tracking aircraft movements, large evasive actions can be avoided. This thesis focuses on the use of stochastic differential equations to enable the effects of wind to be modelled, in order to determine estimates of the probability of conflict. Traffic models have been developed for small networks of aircraft, flying in the presence of wind, to devise conflict resolution strategies, which minimise disruption to the network of aircraft.
H Y Huang (PhD 2003-present)
The Use of MEMS Actuators for Multiple Redundant Active Control of Aircraft
Advances in Microelectronic Mechanical Systems (MEMS) provide possible active flow control schemes which allow designers to replace conventional flight control surfaces. This research programme investigates the use of inflatable binary MEMS actuators which can be used for flow control in all three axes. A detailed CFD study of the aerodynamic properties of MEMS actuators was undertaken to determine the lift and drag properties of individual actuators. An F-16 real-time flight model was adapted to simulate MEMS actuations rather than conventional control surface to demonstrate the potential for MEMS actuations. By providing several hundred MEMS actuators, a high degree of redundancy is possible but the selection of appropriate patterns of actuators, in the presence of failures, is demanding. The research concludes with a simulation to show that the aircraft is controllable in the presence of over 10 failures and that the actuator control patterns can be selected in real-time.
Irfan Madani (PhD 2004-present)
Optimisation of Routing Through Rugged Terrain
In Terrain Reference Navigation (TRN) systems, the aircraft navigation computer selects a route through the terrain by accessing an on-board Digital Terrain Elevation Database (DTED). In practice many routes may provide safe routing through the terrain and a higher level of optimisation is applied to extract the shortest route, to minimise exposure to a threat or to maximise access to a datalink. For military aircraft, one further criteria is pilot work load. This thesis covers the development of a TRN system for real-time routing for military aircraft and the provision of route selection and flight guidance to enable the flight crew to trade-off work load with other mission parameters. Flight guidance is provided in a HUD for an F-16 aircraft. The thesis will conclude with studies to compare predicted work load factors with both perceived and measured pilot work load.
Sen Lu (PhD 2006-present)
Real-time Feature Extraction for a Radar Sensor
This research follows on from previous research in synthetic vision for low-visibility approaches. Using the real-time radar model previously developed, algorithms have been developed for real-time tracking of the runway in a cluttered radar image two to three miles from touchdown. By using the estimate of aircraft position and attitude and knowledge of the runway location, the image is searched for the expected runway shape using area correlation methods. Frame-to-frame coherence is used to predict the implied motion vectors of each runway vertex in the image space in order to reduce the search area and search time. A successful real-time implementation of the tracker has been demonstrated, updating at 20 Hz on a PC.