Articles de revista
http://hdl.handle.net/2117/2253
20180220T14:10:13Z

Nonlinear moving horizon estimator for online estimation of the density and viscosity of a mineral slurry
http://hdl.handle.net/2117/113338
Nonlinear moving horizon estimator for online estimation of the density and viscosity of a mineral slurry
Diaz Castañeda, Jenny Lorena; OcampoMartínez, Carlos; Alvarez, Hernan
This paper proposes a moving horizon estimator for nonlinear systems with unknown inputs, which do not comply with the model structures proposed in the literature for the design of nonlinear observers. The estimator is designed as an optimization problem over a moving horizon, constrained to process model equations and considering the unknown inputs as random inputs among their operating bounds. This proposal is applied to the transport of mineral slurries among process units, typically present in chemical and biological processes. There, to have the slurry properties as online measurements is vital to an efficient control of those processing units. The performance of the proposed estimator is evaluated by simulation with data from a real processing plant, and its performance is compared with a linear estimator executing the same estimation task. Better results are obtained using the proposed estimator by considering the nonlinearities of the process.
20180129T13:37:28Z
Diaz Castañeda, Jenny Lorena
OcampoMartínez, Carlos
Alvarez, Hernan
This paper proposes a moving horizon estimator for nonlinear systems with unknown inputs, which do not comply with the model structures proposed in the literature for the design of nonlinear observers. The estimator is designed as an optimization problem over a moving horizon, constrained to process model equations and considering the unknown inputs as random inputs among their operating bounds. This proposal is applied to the transport of mineral slurries among process units, typically present in chemical and biological processes. There, to have the slurry properties as online measurements is vital to an efficient control of those processing units. The performance of the proposed estimator is evaluated by simulation with data from a real processing plant, and its performance is compared with a linear estimator executing the same estimation task. Better results are obtained using the proposed estimator by considering the nonlinearities of the process.

Relational reinforcement learning for planning with exogenous effects
http://hdl.handle.net/2117/113085
Relational reinforcement learning for planning with exogenous effects
Martinez Martinez, David; Alenyà Ribas, Guillem; Ribeiro, Tony; Inoue, Katsumi; Torras, Carme
Probabilistic planners have improved recently to the point that they can solve difficult tasks with complex and expressive models. In contrast, learners cannot tackle yet the expressive models that planners do, which forces complex models to be mostly handcrafted. We propose a new learning approach that can learn relational probabilistic models with both action effects and exogenous effects. The proposed learning approach combines a multivalued variant of inductive logic programming for the generation of candidate models, with an optimization method to select the best set of planning operators to model a problem. We also show how to combine this learner with reinforcement learning algorithms to solve complete problems. Finally, experimental validation is provided that shows improvements over previous work in both simulation and a robotic task. The robotic task involves a dynamic scenario with several agents where a manipulator robot has to clear the tableware on a table. We show that the exogenous effects learned by our approach allowed the robot to clear the table in a more efficient way.
20180122T21:27:47Z
Martinez Martinez, David
Alenyà Ribas, Guillem
Ribeiro, Tony
Inoue, Katsumi
Torras, Carme
Probabilistic planners have improved recently to the point that they can solve difficult tasks with complex and expressive models. In contrast, learners cannot tackle yet the expressive models that planners do, which forces complex models to be mostly handcrafted. We propose a new learning approach that can learn relational probabilistic models with both action effects and exogenous effects. The proposed learning approach combines a multivalued variant of inductive logic programming for the generation of candidate models, with an optimization method to select the best set of planning operators to model a problem. We also show how to combine this learner with reinforcement learning algorithms to solve complete problems. Finally, experimental validation is provided that shows improvements over previous work in both simulation and a robotic task. The robotic task involves a dynamic scenario with several agents where a manipulator robot has to clear the tableware on a table. We show that the exogenous effects learned by our approach allowed the robot to clear the table in a more efficient way.

Relational reinforcement learning with guided demonstrations
http://hdl.handle.net/2117/113084
Relational reinforcement learning with guided demonstrations
Martínez Martínez, David; Alenyà Ribas, Guillem; Torras, Carme
Modelbased reinforcement learning is a powerful paradigm for learning tasks in robotics. However, indepth exploration is usually required and the actions have to be known in advance. Thus, we propose a novel algorithm that integrates the option of requesting teacher demonstrations to learn new domains with fewer action executions and no previous knowledge. Demonstrations allow new actions to be learned and they greatly reduce the amount of exploration required, but they are only requested when they are expected to yield a significant improvement because the teacher's time is considered to be more valuable than the robot's time. Moreover, selecting the appropriate action to demonstrate is not an easy task, and thus some guidance is provided to the teacher. The rulebased model is analyzed to determine the parts of the state that may be incomplete, and to provide the teacher with a set of possible problems for which a demonstration is needed. Rule analysis is also used to find better alternative models and to complete subgoals before requesting help, thereby minimizing the number of requested demonstrations. These improvements were demonstrated in a set of experiments, which included domains from the international planning competition and a robotic task. Adding teacher demonstrations and rule analysis reduced the amount of exploration required by up to 60% in some domains, and improved the success ratio by 35% in other domains
© <year>. This manuscript version is made available under the CCBYNCND 4.0 license http://creativecommons.org/licenses/byncnd/4.0
20180122T21:20:26Z
Martínez Martínez, David
Alenyà Ribas, Guillem
Torras, Carme
Modelbased reinforcement learning is a powerful paradigm for learning tasks in robotics. However, indepth exploration is usually required and the actions have to be known in advance. Thus, we propose a novel algorithm that integrates the option of requesting teacher demonstrations to learn new domains with fewer action executions and no previous knowledge. Demonstrations allow new actions to be learned and they greatly reduce the amount of exploration required, but they are only requested when they are expected to yield a significant improvement because the teacher's time is considered to be more valuable than the robot's time. Moreover, selecting the appropriate action to demonstrate is not an easy task, and thus some guidance is provided to the teacher. The rulebased model is analyzed to determine the parts of the state that may be incomplete, and to provide the teacher with a set of possible problems for which a demonstration is needed. Rule analysis is also used to find better alternative models and to complete subgoals before requesting help, thereby minimizing the number of requested demonstrations. These improvements were demonstrated in a set of experiments, which included domains from the international planning competition and a robotic task. Adding teacher demonstrations and rule analysis reduced the amount of exploration required by up to 60% in some domains, and improved the success ratio by 35% in other domains

Visual grasp point localization, classification and state recognition in robotic manipulation of cloth: an overview
http://hdl.handle.net/2117/113083
Visual grasp point localization, classification and state recognition in robotic manipulation of cloth: an overview
Jimenez Schlegl, Pablo
Cloth manipulation by robots is gaining popularity among researchers because of its relevance, mainly (but not only) in domestic and assistive robotics. The required science and technologies begin to be ripe for the challenges posed by the manipulation of soft materials, and many contributions have appeared in the last years. This survey provides a systematic review of existing techniques for the basic perceptual tasks of grasp point localization, state estimation and classification of cloth items, from the perspective of their manipulation by robots. This choice is grounded on the fact that any manipulative action requires to instruct the robot where to grasp, and most garment handling activities depend on the correct recognition of the type to which the particular cloth item belongs and its state. The high inter and intraclass variability of garments, the continuous nature of the possible deformations of cloth and the evident difficulties in predicting their localization and extension on the garment piece are challenges that have encouraged the researchers to provide a plethora of methods to confront such problems, with some promising results. The present review constitutes for the first time an effort in furnishing a structured framework of these works, with the aim of helping future contributors to gain both insight and perspective on the subject
© <year>. This manuscript version is made available under the CCBYNCND 4.0 license http://creativecommons.org/licenses/byncnd/4.0/
20180122T21:08:34Z
Jimenez Schlegl, Pablo
Cloth manipulation by robots is gaining popularity among researchers because of its relevance, mainly (but not only) in domestic and assistive robotics. The required science and technologies begin to be ripe for the challenges posed by the manipulation of soft materials, and many contributions have appeared in the last years. This survey provides a systematic review of existing techniques for the basic perceptual tasks of grasp point localization, state estimation and classification of cloth items, from the perspective of their manipulation by robots. This choice is grounded on the fact that any manipulative action requires to instruct the robot where to grasp, and most garment handling activities depend on the correct recognition of the type to which the particular cloth item belongs and its state. The high inter and intraclass variability of garments, the continuous nature of the possible deformations of cloth and the evident difficulties in predicting their localization and extension on the garment piece are challenges that have encouraged the researchers to provide a plethora of methods to confront such problems, with some promising results. The present review constitutes for the first time an effort in furnishing a structured framework of these works, with the aim of helping future contributors to gain both insight and perspective on the subject

On Cayley's factorization of 4D rotations and applications
http://hdl.handle.net/2117/113067
On Cayley's factorization of 4D rotations and applications
Pérez Gracia, Alba; Thomas, Federico
A 4D rotation can be decomposed into a left and a rightisoclinic rotation. This decomposition, known as Cayley’s factorization of 4D rotations, can be performed using the Elfrinkhof–Rosen method. In this paper, we present a more straightforward alternative approach using the corresponding orthogonal subspaces, for which orthogonal bases can be defined. This yields easy formulations, both in the space of 4×44×4 real orthogonal matrices representing 4D rotations and in the Clifford algebra C4,0,0C4,0,0. Cayley’s factorization has many important applications. It can be used to easily transform rotations represented using matrix algebra to different Clifford algebras. As a practical application of the proposed method, it is shown how Cayley’s factorization can be used to efficiently compute the screw parameters of 3D rigidbody transformations.
The final publication is available at link.springer.com
20180122T17:48:43Z
Pérez Gracia, Alba
Thomas, Federico
A 4D rotation can be decomposed into a left and a rightisoclinic rotation. This decomposition, known as Cayley’s factorization of 4D rotations, can be performed using the Elfrinkhof–Rosen method. In this paper, we present a more straightforward alternative approach using the corresponding orthogonal subspaces, for which orthogonal bases can be defined. This yields easy formulations, both in the space of 4×44×4 real orthogonal matrices representing 4D rotations and in the Clifford algebra C4,0,0C4,0,0. Cayley’s factorization has many important applications. It can be used to easily transform rotations represented using matrix algebra to different Clifford algebras. As a practical application of the proposed method, it is shown how Cayley’s factorization can be used to efficiently compute the screw parameters of 3D rigidbody transformations.

Random clustering ferns for multimodal object recognition
http://hdl.handle.net/2117/112785
Random clustering ferns for multimodal object recognition
Villamizar Vergel, Michael Alejandro; Garrell Zulueta, Anais; Sanfeliu Cortés, Alberto; MorenoNoguer, Francesc
We propose an efficient and robust method for the recognition of objects exhibiting multiple intraclass modes, where each one is associated with a particular object appearance. The proposed method, called random clustering ferns, combines synergically a single and realtime classifier, based on the boosted assembling of extremely randomized trees (ferns), with an unsupervised and probabilistic approach in order to recognize efficiently object instances in images and discover simultaneously the most prominent appearance modes of the object through treestructured visual words. In particular, we use boosted random ferns and probabilistic latent semantic analysis to obtain a discriminative and multimodal classifier that automatically clusters the response of its randomized trees in function of the visual object appearance. The proposed method is validated extensively in synthetic and real experiments, showing that the method is capable of detecting objects with diverse and complex appearance distributions in realtime performance.
The final publication is available at link.springer.com
20180115T09:18:59Z
Villamizar Vergel, Michael Alejandro
Garrell Zulueta, Anais
Sanfeliu Cortés, Alberto
MorenoNoguer, Francesc
We propose an efficient and robust method for the recognition of objects exhibiting multiple intraclass modes, where each one is associated with a particular object appearance. The proposed method, called random clustering ferns, combines synergically a single and realtime classifier, based on the boosted assembling of extremely randomized trees (ferns), with an unsupervised and probabilistic approach in order to recognize efficiently object instances in images and discover simultaneously the most prominent appearance modes of the object through treestructured visual words. In particular, we use boosted random ferns and probabilistic latent semantic analysis to obtain a discriminative and multimodal classifier that automatically clusters the response of its randomized trees in function of the visual object appearance. The proposed method is validated extensively in synthetic and real experiments, showing that the method is capable of detecting objects with diverse and complex appearance distributions in realtime performance.

Periodic nonlinear economic model predictive control with changing horizon for water distribution networks
http://hdl.handle.net/2117/112557
Periodic nonlinear economic model predictive control with changing horizon for water distribution networks
Wang, Ye; Salvador Ramon, Jose; Puig Cayuela, Vicenç; Cembrano Gennari, Gabriela
A periodic nonlinear economic model predictive control (EMPC) with changing prediction horizon is proposed for the optimal management of water distribution networks (WDNs). The control model of the WDN is built by means of nonlinear differentialalgebraic equations in which both the hydraulic pressure and flow variables are taken into account. The model allows the controller to consider minimum pressure constraints at the demands. A periodic terminal constraint is employed in order to guarantee closedloop stability. The prediction horizon is modified online in order to guarantee convergence to the optimal periodic trajectory. The proposed control strategy is verified with the case study of the Richmond water network in a realistic hydraulic simulator. Although there are modeling errors between the control model and hydraulic model, the closedloop system converges to a suboptimal periodic trajectory satisfying all the constraints.
20180110T08:01:53Z
Wang, Ye
Salvador Ramon, Jose
Puig Cayuela, Vicenç
Cembrano Gennari, Gabriela
A periodic nonlinear economic model predictive control (EMPC) with changing prediction horizon is proposed for the optimal management of water distribution networks (WDNs). The control model of the WDN is built by means of nonlinear differentialalgebraic equations in which both the hydraulic pressure and flow variables are taken into account. The model allows the controller to consider minimum pressure constraints at the demands. A periodic terminal constraint is employed in order to guarantee closedloop stability. The prediction horizon is modified online in order to guarantee convergence to the optimal periodic trajectory. The proposed control strategy is verified with the case study of the Richmond water network in a realistic hydraulic simulator. Although there are modeling errors between the control model and hydraulic model, the closedloop system converges to a suboptimal periodic trajectory satisfying all the constraints.

3D human pose tracking priors using geodesic mixture models
http://hdl.handle.net/2117/112218
3D human pose tracking priors using geodesic mixture models
Simo Serra, Edgar; Torras, Carme; MorenoNoguer, Francesc
We present a novel approach for learning a finite mixture model on a Riemannian manifold in which Euclidean metrics are not applicable and one needs to resort to geodesic distances consistent with the manifold geometry. For this purpose, we draw inspiration on a variant of the expectationmaximization algorithm, that uses a minimum message length criterion to automatically estimate the optimal number of components from multivariate data lying on an Euclidean space. In order to use this approach on Riemannian manifolds, we propose a formulation in which each component is defined on a different tangent space, thus avoiding the problems associated with the loss of accuracy produced when linearizing the manifold with a single tangent space. Our approach can be applied to any type of manifold for which it is possible to estimate its tangent space. Additionally, we consider using shrinkage covariance estimation to improve the robustness of the method, especially when dealing with very sparsely distributed samples. We evaluate the approach on a number of situations, going from data clustering on manifolds to combining pose and kinematics of articulated bodies for 3D human pose tracking. In all cases, we demonstrate remarkable improvement compared to several chosen baselines.
The final publication is available at link.springer.com
20171218T08:21:41Z
Simo Serra, Edgar
Torras, Carme
MorenoNoguer, Francesc
We present a novel approach for learning a finite mixture model on a Riemannian manifold in which Euclidean metrics are not applicable and one needs to resort to geodesic distances consistent with the manifold geometry. For this purpose, we draw inspiration on a variant of the expectationmaximization algorithm, that uses a minimum message length criterion to automatically estimate the optimal number of components from multivariate data lying on an Euclidean space. In order to use this approach on Riemannian manifolds, we propose a formulation in which each component is defined on a different tangent space, thus avoiding the problems associated with the loss of accuracy produced when linearizing the manifold with a single tangent space. Our approach can be applied to any type of manifold for which it is possible to estimate its tangent space. Additionally, we consider using shrinkage covariance estimation to improve the robustness of the method, especially when dealing with very sparsely distributed samples. We evaluate the approach on a number of situations, going from data clustering on manifolds to combining pose and kinematics of articulated bodies for 3D human pose tracking. In all cases, we demonstrate remarkable improvement compared to several chosen baselines.

Zonotopic fault estimation filter design for discretetime descriptor systems
http://hdl.handle.net/2117/111333
Zonotopic fault estimation filter design for discretetime descriptor systems
Wang, Ye; Wang, Zhenhua; Puig Cayuela, Vicenç; Cembrano Gennari, Gabriela
This paper considers actuatorfault estimation for discretetime descriptor systems with unknown but bounded system disturbance and measurement noise. A zonotopic fault estimation filter is designed based on the analysis of fault detectability indexes. To ensure estimation accuracy, the filter gain in the zonotopic fault estimation filter is optimized through the zonotope minimization. The designed zonotopic filter not only can estimate fault magnitudes, but it also provides fault estimation results in an interval, i.e. the upper and lower bounds of fault magnitudes. Moreover, the proposed fault estimation filter has a nonsingular structure and hence is easy to implement. Finally, simulation results are provided to illustrate the effectiveness of the proposed method.
20171129T12:56:31Z
Wang, Ye
Wang, Zhenhua
Puig Cayuela, Vicenç
Cembrano Gennari, Gabriela
This paper considers actuatorfault estimation for discretetime descriptor systems with unknown but bounded system disturbance and measurement noise. A zonotopic fault estimation filter is designed based on the analysis of fault detectability indexes. To ensure estimation accuracy, the filter gain in the zonotopic fault estimation filter is optimized through the zonotope minimization. The designed zonotopic filter not only can estimate fault magnitudes, but it also provides fault estimation results in an interval, i.e. the upper and lower bounds of fault magnitudes. Moreover, the proposed fault estimation filter has a nonsingular structure and hence is easy to implement. Finally, simulation results are provided to illustrate the effectiveness of the proposed method.

Robust optimization based energy dispatch in smart grids considering simultaneously multiple uncertainties: load demands and energy prices
http://hdl.handle.net/2117/111331
Robust optimization based energy dispatch in smart grids considering simultaneously multiple uncertainties: load demands and energy prices
Nassourou, M; Puig Cayuela, Vicenç; Blesa Izquierdo, Joaquim
Solving the problem of energy dispatch in a heterogeneous complex system is not a trivial task. The problem becomes even more complex considering uncertainties in demands and energy prices. This paper discusses the development of several Economic Model Predictive Control (EMPC) based strategies for solving an energy dispatch problem in a smart microgrid. The smart grid components are described using controloriented model approach. Considering uncertainty of load demands and energy prices simultaneously, and using an economic objective function, leads to a nonlinear nonconvex problem. The technique of using an affine dependent controller is used to convexify the problem. The goal of this research is the development of a controller based on EMPC strategies that tackles both endogenous and exogenous uncertainties, in order to minimize economic costs and guarantee service reliability of the system. The developed strategies have been applied to a hybrid system comprising some photovoltaic (PV) panels, a wind generator, a hydroelectric generator, a diesel generator, and some storage devices interconnected via a DC Bus. Additionally, a comparison between the standard EMPC, and its combination with MPC tracking in singlelayer and twolayer approaches was also carried out based on the daily cost of energy production.
20171129T12:34:38Z
Nassourou, M
Puig Cayuela, Vicenç
Blesa Izquierdo, Joaquim
Solving the problem of energy dispatch in a heterogeneous complex system is not a trivial task. The problem becomes even more complex considering uncertainties in demands and energy prices. This paper discusses the development of several Economic Model Predictive Control (EMPC) based strategies for solving an energy dispatch problem in a smart microgrid. The smart grid components are described using controloriented model approach. Considering uncertainty of load demands and energy prices simultaneously, and using an economic objective function, leads to a nonlinear nonconvex problem. The technique of using an affine dependent controller is used to convexify the problem. The goal of this research is the development of a controller based on EMPC strategies that tackles both endogenous and exogenous uncertainties, in order to minimize economic costs and guarantee service reliability of the system. The developed strategies have been applied to a hybrid system comprising some photovoltaic (PV) panels, a wind generator, a hydroelectric generator, a diesel generator, and some storage devices interconnected via a DC Bus. Additionally, a comparison between the standard EMPC, and its combination with MPC tracking in singlelayer and twolayer approaches was also carried out based on the daily cost of energy production.