Home > Research > Publications & Outputs > Event-Triggered Based State Estimation for Auto...

Electronic data

View graph of relations

Event-Triggered Based State Estimation for Autonomous Operation of an Aerial Robotic Vehicle

Research output: Contribution to journalJournal article

Forthcoming
<mark>Journal publication date</mark>20/02/2019
<mark>Journal</mark>IFAC-PapersOnLine
Publication statusAccepted/In press
Original languageEnglish

Abstract

In this article the problem of event-triggered (ET) state estimation for autonomous navigation of an aerial vehicle is investigated numerically. The aerial vehicle is considered as a general example of a nonlinear non-Gaussian system for state estimation under process and measurement noise. The motivation behind the problem is the conditions that the aerial vehicles are facing in extreme and hazardous environments due to constant exposure of the sensors and actuators to the high frequency process and measurement noises. Here we consider autonomous operation of a quadcopter for mapping of a radioactive environment, where the quadcopter may subject to radiations and non-Gaussian noises. Autonomous operation of the aerial vehicle with a limited available energy and for a longer period of time, demands an efficient management of the energy sources. Therefore, in this study we take the first step towards this goal by studying an event triggering strategy in which the data measured by the sensors is transmitted to the processing unit only if certain events happen. The sensor employed for navigation purpose is the inertial measurement unit, including accelerometers and gyroscopes, used to estimate the quadcopter states only when their measurements are informative. An event-triggered particle filtering (PF) state estimation technique is adopted for this application. The choice of particle filter as state-estimator is inevitable not only because of nonlinear and non-Gaussian nature of the system, but also because of non-Gaussianity of the conditional distribution of the posteriori probability density function resulting from the event triggering. In the proposed method, it is proved that particles are weighted differently in the case of event triggering and no triggering. The numerical results for robust nonlinear attitude stabilization of the quadcopter in the presence of event-triggered particle filter state estimation confirm the efficiency of the proposed method.