Search

In this paper, an established distributed LQR control methodology applied to identical linear systems is extended to control arbitrary formations of non-identical UAV's. The nonlinear model of a low-speed experimental UAV known as X-RAE1 is utilized for simulation purposes. The formation is composed of four dynamically decoupled X-RAE1 which differ in their masses and their products of inertia about the xz plane. In order to design linear controllers the nonlinear models are linearized for horizontal flight conditions at constant velocity. State-feedback, input and similarity transformations are applied to solve model-matching type problems and compensate for the mismatch in the linearized models due to mass and symmetry discrepancies among the X-RAE1 models. It is shown that the method is based on the controllability indices of the linearized models. Distributed LQR control employed in networks of identical linear systems is appropriately adjusted and applied to the formation of the nonidentical UAV's. The applicability of the approach is illustrated via numerous simulation results.

Cycling is an increasingly popular mode of travel in cities due to the great advantages that it offers in terms of space consumption, health and environmental sustainability, and is therefore favoured and promoted by many city authorities. However, the relatively low perceived safety of cycling from the users’ side currently presents itself as a hurdle towards higher uptake levels of cycling, and unfortunately, road accident statistics (1) confirm this perception as reality. A typical collision pattern observed involves cyclists being “crushed” by turning motorised vehicles, due to their presence in the so-called “blind spot”, which is to the left of the vehicle in the UK and to the right in countries with right - hand traffic (2). Up until a few years ago, th e only options for tackling such a problem would be drawn from the domain of “hard” traffic engineering measures, (usually cost-intensive and/or severely disruptive, such as segregated lanes or vehicle type bans in certain streets). However, trends in the development of ubiquitous computing now offer smaller, more accurate and durable tools to support traffic safety interventions. Examples range from simple passive measures (3) to more advanced experimental active cyclist detection system (4). But while such solutions certainly represent steps in the right direction in terms of preventing cyclist - vehicle collisions, they are limited in what they are unable to perform any reliable prediction of accidents due to their inability to accurately track the cyclist’s trajectory and estimate his/her position in a critical time-horizon of 5-10 seconds. Indeed, the accurate (< 1 m) localisation of the cyclist is a necessity when it comes to preventing collisions, but so far remains an important unresolved challenge, as none of the existing mainstream technologies (GPS, WiFi etc.) can achieve it. Enhanced positioning systems, on the other hand, such as U-blox (5) and Spatial (6) Inertial Navigation System (INS), can achieve accurate positioning in theory, but they are specifically designed for four-wheel vehicles and are therefore very expensive when used for tracking bicycles. Besides, the dynamics of a bicycle is very complex and different from an ordinary vehicle, and so the accuracy of such enhanced positioning syst ems will differ greatly when used on a bicycle. The research reported here focuses on the development and testing of an innovative technological solution for accurately localising and tracking cyclists in urban environments using a low-cost micro-electrome chanical systems (MEMS) sensor configuration on a prototype instrumented bicycle system, called “ iBike ” (7). The ultimate goal is to develop a collision prediction and avoidance system, and the present paper presents a novel fusion technique that could be utilised to improve localisation accuracy based on Wireless Communication Technologies (WCT) widely found in cities as well as Global Navigation Satellite System (GNSS) positioning.

The paper aims at formulating an integrated approach for the selection of decentralized control structures using a number of structural criteria aiming at facilitating the design of decentralised control schemes. This requires the selection of decentralisation structure that will allow the generic solvability of a variety of decentralised control problems, such as pole assignment by decentralised output feedback. The approach is based on the use of necessary and sufficient conditions for generic solvability and exact solvability of decentralised control problems. The generic solvability conditions lead to characterisations of inputs and outputs channel partitions. The exact solvability conditions use criteria on avoiding the presence of fixed modes, as well as necessary conditions for pole assignment, expressed in terms of properties of Plucker invariants and Markov type matrices. The structural approach provides a classification of desirable input and output partitions based on structural criteria and it is embedded in an overall framework that may involve aspects related to large scale design. (C) 2016, IFAC (International Federation of Automatic Control) Hosting by Elsevier Ltd. All rights reserved.

A six-degrees of freedom mathematical model of an experimental Remotely Piloted Vehicle (RPV) and the linearised longitudinal and lateral models at 30 m/sec are developed. The longitudinal and lateral dynamics are analysed and the equivalent discrete systems are used to provide baseline data for the identification of the aerodynamic derivatives of the RPV. An advanced aircraft parameter estimation method - the Extended Kalman F ilte r - is implemented for the estimation of the aerodynamic characteristics of the RPV. Conclusions are drawn about the id e n tifi- a b ility of the s tab ility and control derivatives from pitch, roll and yaw rate measurements. MPhil