Le concept d’habitat intelligent s’est largement développé ces dernières années afin de proposer des solutions face à deux préoccupations majeures : la gestion optimisée de l’énergie dans le bâtiment et l’aide au maintien à domicile de personnes âgées. C’est dans ce contexte que le projet CAPTHOM, dans lequel s’inscrit cette thèse, a été développé. Pour répondre à ces problématiques, de nombreux capteurs, de natures différentes, sont utilisés pour la détection de la présence humaine, la détermination de la localisation et de la posture de la personne. En effet, aucun capteur, ne peut, seul, répondre à l’ensemble de ces informations justifiant le développement d’un dispositif multi-capteurs et d’une politique de fusion de données. Dans ce projet, les capteurs retenus sont les détecteurs infrarouges passifs, les thermopiles et la caméra. Aucun capteur n’est porté par la personne (non invasivité du dispositif). Nous proposons une architecture globale du capteur intelligent composée de quatre modules de fusion permettant respectivement de détecter la présence humaine, de localiser en 3D la personne, de déterminer la posture et d’aider à la prise de décision finale selon l’application visée. Le module de détection de présence fusionne les informations des trois capteurs : les détecteurs IRP pour la détection du mouvement, les thermopiles pour la présence en cas d’immobilité de la personne et la caméra pour identifier l’entité détectée. La localisation 3D de la personne est réalisée grâce à l’estimation de position sur horizon glissant. Cette méthode, nommée Visual Receding Horizon Estimation (VRHE), formule le problème d’estimation de position en un problème d’optimisation non linéaire sous contraintes dans le plan image. Le module de fusion pour la détermination de posture s’appuie sur la théorie des ensembles flous. Il assure la détermination de la posture indépendamment de la personne et de sa distance vis à vis de la caméra. Enfin, un module d’aide à la décision fusionne les sorties des différents modules et permet de déclencher des alarmes dans le cas de la surveillance de personnes âgées ou de déclencher des applications domotiques (chauffage, éclairage) pour la gestion énergétique de bâtiments. / The smart home concept has been widely developed in the last years in order to propose solutions for twomain concerns : optimized energy management in building and help for in-home support for elderly people.In this context, the CAPTHOM project, in which this thesis is in line with, has been developed. To respondto these problems, many sensors, of different natures, are used to detect the human presence, to determinethe position and the posture of the person. In fact, no sensor can , alone, answers to all information justifyingthe development of a multi-sensor system and a data fusion method. In this project, the selected sensorsare passive infrared sensors (PIR), thermopiles and a video camera. No sensor is carried by the person(non invasive system). We propose a global architecture of intelligent sensor made of four fusion modulesallowing respectively to detect the human presence, to locate in 3D the person, to determine the posture andto help to make a decision according to the application. The human presence module fuses information ofthe three sensors : PIR sensors for the movement, thermopiles for the presence in case of immobility and thecamera to identify the detected entity. The 3D localisation of the person is realized thanks to position recedinghorizon estimation. This method, called Visual Receding Horizon Estimation (VRHE), formulates the positionestimation problem into an nonlinear optimisation problem under constraints in the image plane. The fusionmodule for the posture determination is based on fuzzy logic. It insures the posture determination regardlessof the person and the distance from the camera. Finally, the module to make a decision fuses the outputs of the preceding modules and gives the opportunity to launch alarms (elderly people monitoring) or to commandhome automation devices (lightning, heating) for the energy management of buildings.

We present a new trajectory generation algorithm for autonomous guidance and control of unmanned vehicles from a given starting point to a given target location. We build and update incomplete a priori maps of the operating environment in real time using onboard sensors and compute level sets on the map reflecting the minimal cost of traversal from the current vehicle location to the goal. We convert the trajectory generation problem into a finite-time-horizon optimal control problem using the computed level sets as terminal costs in a receding horizon framework and transform it into a simpler nonlinear programming problem by discretization of the candidate control and state histories. We ensure feasibility of the generated trajectories by constraining the solution of the optimization problem using a simplified vehicle model. We provide strong performance guarantees by checking for stability of the algorithm through the test of matching conditions at the end of each iteration. The algorithm thus explicitly incorporates the vehicle dynamics and constraints and generates trajectories realizable by the vehicle in the field. Successful preliminary field demonstrations and complete simulation results for a marine unmanned surface vehicle demonstrate the efficacy of the proposed approach for fast operations in poorly characterized riverine environments. / Master of Science

No description available.

Scheduling refinery operation promises a big cut in logistics cost, maximizes efficiency, organizes allocation of material and resources, and ensures that production meets targets set by planning team. Obtaining accurate and reliable schedules for execution in refinery plants under different scenarios has been a serious challenge. This research was undertaken with the aim to develop robust methodologies and solution procedures to address refinery scheduling problems with uncertainties in process parameters. The research goal was achieved by first developing a methodology for short-term crude oil unloading and transfer, as an extension to a scheduling model reported by Lee et al. (1996). The extended model considers real life technical issues not captured in the original model and has shown to be more reliable through case studies. Uncertainties due to disruptive events and low inventory at the end of scheduling horizon were addressed. With the extended model, crude oil scheduling problem was formulated under receding horizon control framework to address demand uncertainty. This work proposed a strategy called fixed end horizon whose efficiency in terms of performance was investigated and found out to be better in comparison with an existing approach. In the main refinery production area, a novel scheduling model was developed. A large scale refinery problem was used as a case study to test the model with scheduling horizon discretized into a number of time periods of variable length. An equivalent formulation with equal interval lengths was also presented and compared with the variable length formulation. The results obtained clearly show the advantage of using variable timing. A methodology under self-optimizing control (SOC) framework was then developed to address uncertainty in problems involving mixed integer formulation. Through case study and scenarios, the approach has proven to be efficient in dealing with uncertainty in crude oil composition.

A novel application is presented for a fault-tolerant adaptive model predictive control system for small-scale helicopters. The use of a joint Extended Kalman Filter, (EKF), for the estimation of the states and parameters of the UAV, provided the advantage of implementation simplicity and accuracy. A linear model of a small-scale helicopter was utilized for testing the proposed control system. The results, obtained through the simulation of different fault scenarios, demonstrated that the proposed scheme was able to handle different types of actuator and system faults effectively. The types of faults considered were represented in the parameters of the mathematical representation of the helicopter. Benefits provided by the proposed fault-tolerant adaptive model predictive control systems include: The use of the joint Kalman filter provided a straightforward approach to detect and handle different types of actuator and system faults, which were represented as changes of the system and input matrices. The built-in adaptability provided for the handling of slow time-varying faults, which are difficult to detect using the standard residual approach. The successful inclusion of fault tolerance yielded a significant increase in the reliability of the UAV under study. A byproduct of this research is an original comparison between the EKF and the Unscented Kalman Filter, (UKF). This comparison attempted to settle the conflicting claims found in the research literature concerning the performance improvements provided by the UKF. The results of the comparison indicated that the performance of the filters depends on the approximation used for the nonlinear model of the system. Noise sensitivity was found to be higher for the UKF, than the EKF. An advantage of the UKF appears to be a slightly faster convergence.

Scheduling refinery operation promises a big cut in logistics cost, maximizes efficiency, organizes allocation of material and resources, and ensures that production meets targets set by planning team. Obtaining accurate and reliable schedules for execution in refinery plants under different scenarios has been a serious challenge. This research was undertaken with the aim to develop robust methodologies and solution procedures to address refinery scheduling problems with uncertainties in process parameters. The research goal was achieved by first developing a methodology for short-term crude oil unloading and transfer, as an extension to a scheduling model reported by Lee et al. (1996). The extended model considers real life technical issues not captured in the original model and has shown to be more reliable through case studies. Uncertainties due to disruptive events and low inventory at the end of scheduling horizon were addressed. With the extended model, crude oil scheduling problem was formulated under receding horizon control framework to address demand uncertainty. This work proposed a strategy called fixed end horizon whose efficiency in terms of performance was investigated and found out to be better in comparison with an existing approach. In the main refinery production area, a novel scheduling model was developed. A large scale refinery problem was used as a case study to test the model with scheduling horizon discretized into a number of time periods of variable length. An equivalent formulation with equal interval lengths was also presented and compared with the variable length formulation. The results obtained clearly show the advantage of using variable timing. A methodology under self-optimizing control (SOC) framework was then developed to address uncertainty in problems involving mixed integer formulation. Through case study and scenarios, the approach has proven to be efficient in dealing with uncertainty in crude oil composition.

Covariance assignment theory, introduced in the late 1980s, provided the only means to directly control the steady-state error properties of a linear system subject to Gaussian white noise and parameter uncertainty. This theory, however, does not extend to control of the transient uncertainties and to date there exist no practical engineering solutions to the problem of directly and optimally controlling the uncertainty in a linear system from one Gaussian distribution to another. In this thesis I design a dual-mode Receding Horizon Controller (RHC) that takes a controllable, deterministic linear system from an arbitrary initial covariance to near a desired stationary covariance in finite time. The RHC solves a sequence of free-time Optimal Control Problems (OCP) that directly controls the fundamental solution matrices of the linear system; each problem is a right-invariant OCP on the matrix Lie group GLn of invertible matrices. A terminal constraint ensures that each OCP takes the system to the desired covariance. I show that, by reducing the Hamiltonian system of each OCP from T?GLn to gln? x GLn, the transversality condition corresponding to the terminal constraint simplifies the two-point Boundary Value Problem (BVP) to a single unknown in the initial or final value of the costate in gln?. These results are applied in the design of a dual-mode RHC. The first mode repeatedly solves the OCPs until the optimal time for the system to reach the de- sired covariance is less than the RHC update time. This triggers the second mode, which applies covariance assignment theory to stabilize the system near the desired covariance. The dual-mode controller is illustrated on a planar system. The BVPs are solved using an indirect shooting method that numerically integrates the fundamental solutions on R4 using an adaptive Runge-Kutta method. I contend that extension of the results of this thesis to higher-dimensional systems using either in- direct or direct methods will require numerical integrators that account for the Lie group structure. I conclude with some remarks on the possible extension of a classic result called Lie?s method of reduction to receding horizon control.

This dissertation addresses path planning for an autonomous vehicle navigating in a two dimensional environment for which an a priori map is inaccurate and for which the environment is sensed in real-time. For this class of application, planning decisions must be made in real-time. This work is motivated by the need for fast autonomous vehicles that require planning algorithms to operate as quickly as possible. In this dissertation, we first study the case in which there are only static obstacles in the environment. We propose a hybrid receding horizon control path planning algorithm that is based on level-set methods. The hybrid method uses global or local level sets in the formulation of the receding horizon control problem. The decision to select a new level set is made based on certain matching conditions that guarantee the optimality of the path. We rigorously prove sufficient conditions that guarantee that the vehicle will converge to the goal as long as a path to the goal exists. We then extend the proposed receding horizon formulation to the case when the environment possesses moving obstacles. Since all of the results in this dissertation are based on level-set methods, we rigorously investigate how level sets change in response to new information locally sensed by a vehicle. The result is a dynamic fast marching algorithm that usually requires significantly less computation that would otherwise be the case. We demonstrate the proposed dynamic fast marching method in a successful field trial for which an autonomous surface vehicle navigated four kilometers through a riverine environment. / Ph. D.

<p>This thesis presents a method for the design of a predictive controller with variable step sizes.Predictive methods such as receding horizon control (or model predictive control) use aa fixed sampling frequency when updating the inputs. In the proposed method, the switchingtimes are incorporated into an optimization problem, thus resulting in anadaptive step-size control process. The controller with variable timesteps is shown to require less tuning and to reduce the number of expensive model evaluations.An alternate solution approach had to be developed to accommodate the new problem formulation.The controller's stability is proven in a context that does not require terminal cost or constraints.The thesis presents examples that compare the performance of the variable switching time controllerwith the receding horizon method with a fixed step size. This research opens many roads for futureextension of the theoretical work and practical applications of the controller.</p> / Doctor of Science (PhD)

This work is inspired by robotic search applications where a robot or team of robots is equipped with sensors and tasked to autonomously acquire as much information as possible from a region of interest. To accomplish this task, robots must plan paths through the region of interest that maximize the effectiveness of the sensors they carry. Receding horizon path planning is a popular approach to addressing the computationally expensive task of planning long paths because it allows robotic agents with limited computational resources to iteratively construct a long path by solving for an optimal short path, traversing a portion of the short path, and repeating the process until a receding horizon path of the desired length has been constructed. However, receding horizon paths do not retain the optimality properties of the short paths from which they are constructed and may perform quite poorly in the context of achieving the robotic search objective. The primary contributions of this work address the worst-case performance of receding horizon paths by developing methods of using terminal rewards in the construction of receding horizon paths. We prove that the proposed methods of constructing receding horizon paths provide theoretical worst-case performance guarantees. Our result can be interpreted as ensuring that the receding horizon path performs no worse in expectation than a given sub-optimal search path. This result is especially practical for subsea applications where, due to use of side-scan sonar in search applications, search paths typically consist of parallel straight lines. Thus for subsea search applications, our approach ensures that expected performance is no worse than the usual subsea search path, and it might be much better. The methods proposed in this work provide desirable lower-bound guarantees for a single robot as well as teams of robots. Significantly, we demonstrate that existing planning algorithms may be easily adapted to use our proposed methods. We present our theoretical guarantees in the context of subsea search applications and demonstrate the utility of our proposed methods through simulation experiments and field trials using real autonomous underwater vehicles (AUVs). We show that our worst-case guarantees may be achieved despite non-idealities such as sub-optimal short-paths used to construct the longer receding horizon path and unreliable communication in multi-agent planning. In addition to theoretical guarantees, An important contribution of this work is to describe specific implementation solutions needed to integrate and implement these ideas for real-time operation on AUVs. / Doctor of Philosophy / This work is inspired by robotic search applications where a robot or team of robots is equipped with sensors and tasked to autonomously acquire as much information as possible from a region of interest. To accomplish this task, robots must plan paths through the region of interest that maximize the effectiveness of the sensors they carry. Receding horizon path planning is a popular approach to addressing the computationally expensive task of planning long paths because it allows robotic agents with limited computational resources to iteratively construct a long path by solving for an optimal short path, traversing a portion of the short path, and repeating the process until a receding horizon path of the desired length has been constructed. However, receding horizon paths do not retain the optimality properties of the short paths from which they are constructed and may perform quite poorly in the context of achieving the robotic search objective. The primary contributions of this work address the worst-case performance of receding horizon paths by developing methods of using terminal rewards in the construction of receding horizon paths. The methods proposed in this work provide desirable lower-bound guarantees for a single robot as well as teams of robots. We present our theoretical guarantees in the context of subsea search applications and demonstrate the utility of our proposed methods through simulation experiments and field trials using real autonomous underwater vehicles (AUVs). In addition to theoretical guarantees, An important contribution of this work is to describe specific implementation solutions needed to integrate and implement these ideas for real-time operation on AUVs.