Da-Wei Gu

State-space formulae for discrete-time H∞ optimization

Abstract State-space formulae and proofs are given for ail the important steps in discrete-time H 8 optimization. The steps closely follow the algorithms for continuous-time systems, but there are some technically involved differences in the detail that make their derivation non-trivial.

An algorithm for super-optimal H ∞ design: the two-block case

Abstract In this paper, we consider the strengthened H∞ optimization problem in the two-block case. An algorithm for the super-optimal solution is proposed. It is based on results by the authors (IEEE Trans. Aut. Control, AC-33, 833 (1988)) for the less practical one-block case. It uses state-space models and is numerically implementable. A feature of the procedure is an optimality criterion which is a natural generalization of that used in the iterative scheme of Doyle (Lecture Notes in Advances in Multivariable Control, ONR/Honeywell Workshop, Minneapolis, U.S.A. (1984)) for the unstrengthened problem.

Design of linear parameter varying trajectory tracking controllers for an unmanned air vehicle

Abstract This article presents a trajectory tracking controller design approach for an unmanned air vehicle using the linear parameter varying (LPV) methods. The longitudinal and lateral controllers are designed using an inner-outer loop structure with the inner loop LPV controller designed first using μ-synthesis. The inner loop is then approximated with a reference model and the outer loop is designed using loop-shaping techniques. Flight trajectory is expressed in terms of earth co-ordinates, which the outer loop controller converts into reference signals for the inner loop. Full-scale non-linear simulations are used to test the efficiency of the designed controller and of the proposed design approach.

Unmanned air vehicle air data estimation using a matrix of pressure sensors: a comparison of neural networks and look-up tables

Flush airdata sensing (FADS) systems are cost- and weight- effective alternatives to current air data booms for measuring important air data parameters such as airspeed, angle of attack, sideslip, etc. Most applications consider large manned/unmanned air vehicles where the Pitot-static tube is located at the nose tip. However, traditional air data booms can be physically impractical for micro- (unmanned) air vehicles (MAVs) and, in this article, a FADS system mounted on the wing leading edge of a MAV flown at low speeds of Mach 0.07 (wind tunnel experiments under corresponding conditions) is designed. Moreover, two approaches for converting the FADS system pressure to meaningful air data are compared: a neural network (NN) approach and a look-up table (LUT). Results have shown that instrumentation weight and cost were reduced by 80 per cent and 97 per cent, respectively, in comparison to a traditional air data boom. Overall, the NN estimation accuracies were 0.51°, 0.44 lb/ft2, and 0.62 m/s and the LUT estimation accuracies 1.32°, 0.11 lb/ft2, and 0.88 m/s for the angle of attack, static pressure, and airspeed, respectively. It was also found that the LUT has faster execution times while the NN was in most cases more robust to sensor faults. However, while the LUT requires high memory usage, especially for higher dimensions, the NN can be executed in a few lines of code.

Enabling the operation of multiple micro-air-vehicles in increasingly complex obstacle-rich environments

This paper considers the guidance and control of unmanned vehicles in complex obstaclerich environments. The complexity of the environment requires that the local obstacle space is continually considered (particularly in the presence of disturbances, dynamic obstacles or other vehicles), and this is aided by splitting the motion planning problem into distinct global and local layers. The primary focus of this paper is on the development of a continuous, dynamically feasible, local motion planning framework, which is applied in simulation to a quadrotor micro air vehicle. This local motion planning layer is effectively an implementation of a receding horizon control problem, using an output space design to enable increased horizon length and faster optimisation. Polynomial functions are used to describe local motion trajectories which are constrained to vehicle performance limits and optimised to track a pre -defined global trajectory while also considering the local obstacle space. Static and dynamic obstacles are considered, as well as the immediate-term decentralised deconfliction of multiple unmanned vehicles, and multiple formations of unmanned vehicles.

Continuous Local Motion Planning & Control for Micro- Air-Vehicles in Complex Environments

This paper considers the guidance and control of unmanned vehicles in complex obstaclerich environments. The need to consider continually the local obstacle space, even when tracking a trajectory that was designed with that s pace in mind, leads to a motion planning architecture with distinct global & local elements. Receding horizon control (RHC) provides a framework to address the continuous local motion planning problem, but typically suffers from the complexity of a control space design, part icularly for a vehicle with fast dynamics. This paper addresses this issue by subdividing the RHC problem into distinct output space (trajectory design) and control space (trajectory t racking) problems. The output space design is implemented as a real-time optimization p roblem, with receding horizon trajectories described by polynomial functions, whi ch can be constrained to vehicle performance limits and sensor confirmed free-space. The control space problem is implemented via multi-loop pure-pursuit style track ing loops with feed-forward demands in each axis. Simulation results based on a nonlinear 6dof model of a quadrotor micro-airvehicle are provided, demonstrating successful handling of the dynamic interaction between the separated output and control space components of the RHC problem.

Situation aware trajectory tracking for micro air vehicles in obstacle-rich environments

When operating unmanned vehicles within complex, obstacle rich, environments there is a need to consider continually the local obstacle space, even when tracking a trajectory designed with that space in mind. Disturbances such as gusts, of particular importance to micro air vehicles, inevitably lead to trajectory tracking errors which in turn require trajectory re-acquire manoeuvres that must be conducted with an awareness of the surrounding obstacle space. Additionally, there will exist a class of unmapped or dynamic obstacles that will require en-route detection, but which can be handled intuitively without impacting on large scale or global plan. The work discussed in this paper is aimed at handling this class of obstacle and disturbance by providing trajectory tracking algorithms with a ‘situation awareness’. This is done by creating a continuous local motion planning layer, that sits between a global (or large scale) planner and the vehicle autopilot. This local motion planning layer is implemented as a constrained optimisation problem, combining: (i) Vehicle Performance Limits, (ii) Local Obstacle Information & (iii) Environmental Conditions, into a receding horizon framework that continuously designs safe and dynamically feasible local trajectories.

Fault Detection and Isolation (FDI)

With the growing use of complex systems, there has been considerable interest in the development of techniques to detect and isolate faults. An undetected fault in a system can have catastrophic effects such as loss of human life, environmental pollution and financial losses. Examples of places where FDI schemes can be useful are hospitals and manufacturing companies. In hospitals, staff would need to be aware of faults in health monitoring equipment (e.g. electrocardiographs) to avoid incorrect patient health diagnosis. On the other hand an undetected fault in a production line can eventually require overall plant shutdown, which can be costly. The literature and effort gone into the field of FDI is overwhelming. It still remains one of the most active areas of research today. Owing to this, one can expect the terminology to be quite misleading as different authors assign different terms to describe similar concepts.

Super-Optimal H∞ Design

In this paper, we examine the usefulness of super-optimal HOO design in robust control. It is argued that the approach can lead to a more robust design than standard H OO control, but that there is usually a cost to pay in complexity of controller. We outline a recently developed state-space approach for solving 2-block problems, which enables realistic designs such as mixed sensitivity to be solved. This is then used to design the super-optimal controller for a simple example and the results are compared with those of a standard H oo controller.

Stable H- an H∞ control-system design package

Stable-H is a CAD package for designing H∞ optimal controllers for linear, time-invariant finite-dimensional continuous-time, multivariable systems. H2 (eg, LQG) optimal controllers can also be designed but the main purpose of Stable-H is H∞ design. The name Stable-H was coined because the package optimises, over the set of all stabilising controllers, the H∞ or H2 norm of a given cost function. In this paper we will describe the important features of Stable-H and give an example of its application.