During the last few decades there has been an extensive study on the design of observers for nonlinear systems. An observer or estimator can be defined as a process that provides in real time the estimate of the state (or some function of it) of the plant from partial and possibly noisy measurements of the inputs and outputs and inexact knowledge of the initial condition. The aim of this project is to Develop Nonlinear Observers (DENO) that are provably accurate by construction. In particular, to assure that the research is driven by high-impact application areas, the DENO project will focus on the following class of nonlinear observers:
Minimum-energy and H-infinity state estimators for systems with implicit outputs: The main practical motivation for the study of systems with implicit outputs is dynamic vision. In particular, the estimation of the position and attitude of an autonomous robotic vehicle with respect to a desired coordinate system, defined by visual landmarks, by using measurements from a camera mounted on-board. This problem is highly nonlinear and poses considerable challenges. Furthermore, the observers must deal with the usual problems associated with vision systems such as noise, latency and intermittency of observations.
Range observers: The main motivation for its study arises in the underwater field. Over the last decade, applications with ocean robotics have increased dramatically. The use of remotely operated vehicles and, more recently, autonomous underwater vehicles have shown to be extremely important tools to study and explore the oceans. A key enabling element for the use of such robotic vehicles is the availability of advance navigation and positioning systems. Since electromagnetic signals do not penetrate below the sea surface, a solution for communication and positioning is to use acoustic systems. Typically, these systems only provide range measurements and are plagued with intermittent failures, latency, and multi-path effects.
Multi-model adaptive estimators: The aim is to design adaptive observers (joint state estimation and parameter identification) with special emphasis on methodologies that use multiple-model architectures. Multi-model estimators integrate dynamic hypotheses testing concepts with linear or nonlinear observers. They are at the heart of modern surveillance systems involving multiple sensors and multiple objects. Other motivations for the study of this type of observers are robust adaptive control and fault detection and isolation, including the reconfiguration of the control system.
State estimators of networked systems: The last few years have witnessed considerably effort on the deployment of networked sensors distributed over a wide area. Much of this interest has been motivated by applications in a variety of areas that range from environmental monitoring and surveillance, to navigation of a moving vehicle. However, the presence of a network poses a number of challenging problems because the communication network itself is a dynamical system that exhibits characteristics that traditionally have not been taken into account in observer system design. These special characteristics include communication delays and data loss across the network. Thus, the problem of estimating the state of a remote system based on measurements carried through a network must be addressed taking into account the communication channel.