The novel paradigm of collaborative automation, with machines and industrial robots that synergically share the same workspace with human workers, requires to rethink how activities are prioritized in order to account for possible variabilities in their durations. This article proposes a scheduling method for collaborative assembly tasks that allows to optimally plan assembly activities based on the knowledge acquired during runtime and so adapts to variations along the life cycle of a manufacturing process. The scheduler is based on time Petri nets and the output plan is optimized by minimizing the idle time of each agent. The experimental validation carried out on a realistic industrial use-case consisting of a small assembly line with two robots and a human operator confirms the effectiveness of the approach.

DNA microarray datasets are characterized by a large number of features with very few samples, which is a typical cause of overfitting and poor generalization in the classification task. Here, we introduce a novel feature selection (FS) approach which employs the distance correlation (dCor) as a criterion for evaluating the dependence of the class on a given feature subset. The dCor index provides a reliable dependence measure among random vectors of arbitrary dimension, without any assumption on their distribution. Moreover, it is sensitive to the presence of redundant terms. The proposed FS method is based on a probabilistic representation of the feature subset model, which is progressively refined by a repeated process of model extraction and evaluation. A key element of the approach is a distributed optimization scheme based on a vertical partitioning of the dataset, which alleviates the negative effects of its unbalanced dimensions. The proposed method has been tested on several microarray datasets, resulting in quite compact and accurate models obtained at a reasonable computational cost.

It is widely agreed that future manufacturing environments will be populated by humans and robots sharing the same workspace. However, the real collaboration can be sporadic, especially in the case of assembly tasks, which might involve autonomous operations to be executed by either the robot or the human worker. In this scenario, it might be beneficial to predict the actions of the human in order to control the robot both safely and efficiently. In this paper, we propose a method to predict human activity patterns in order to early infer when a specific collaborative operation will be requested by the human and to allow the robot to perform alternative autonomous tasks in the meanwhile. The prediction algorithm is based on higher-order Markov chains and is experimentally verified in a realistic scenario involving a dual-arm robot employed in a small part collaborative assembly task.

We here introduce a novel classification approach adopted from the nonlinear model identification framework, which jointly addresses the feature selection (FS) and classifier design tasks. The classifier is constructed as a polynomial expansion of the original features and a selection process is applied to find the relevant model terms. The selection method progressively refines a probability distribution defined on the model structure space, by extracting sample models from the current distribution and using the aggregate information obtained from the evaluation of the population of models to reinforce the probability of extracting the most important terms. To reduce the initial search space, distance correlation filtering is optionally applied as a preprocessing technique. The proposed method is compared to other well-known FS and classification methods on standard benchmark problems. Besides the favorable properties of the method regarding classification accuracy, the obtained models have a simple structure, easily amenable to interpretation and analysis.

This paper presents a model predictive control approach for the economic optimization of a microgrid including smart buildings, wind power production facilities and an energy storage unit. Various optimization scenarios are considered in a comprehensive and unified framework, which can be adapted to pursue different objectives at the same time, such as ensuring the electricity supply to the smart buildings, maximizing the profit from the electricity trading market, or managing the energy storage. The optimization problem can be addressed in a model predictive framework using the receding horizon approach, and ultimately formulated as a quadratic programming problem, which can be solved with reliable and efficient tools. In order to analyze a realistic scenario, the relevant data are taken from real systems (i.e., from a real wind farm and from a real commercial building, located in Italy). Simulation results show the economic advantages that can be gained through the combined usage of renewable energy generation and energy storage.

This paper concerns the design of an energy management system for a building cooling system that includes a chiller plant (with two or more chiller units), a thermal storage unit, and a cooling load. The latter is modeled in a probabilistic framework to account for the uncertainty in the building occupancy. The energy management task essentially consists in the minimization of the energy consumption of the cooling system, while preserving comfort in the building. This is achieved by a twofold strategy. The cooling power request is optimally distributed among the chillers and the thermal storage unit. At the same time, a slight modulation of the temperature set-point of the zone is allowed, trading energy saving for comfort. The problem can be decoupled into a static optimization problem (mainly addressing the chiller plant optimization) and a dynamic programming (DP) problem for a discrete time stochastic hybrid system (SHS) that takes care of the overall energy minimization. The DP problem is solved by abstracting the SHS to a (finite) controlled Markov chain, where costs associated with state transitions are computed by simulating the original model and determining the corresponding energy consumption. A numerical example shows the efficacy of the approach.

Classical approximate dynamic programming techniques based on state-space gridding become computationally impracticable for high-dimensional problems. Policy search techniques cope with this curse of dimensionality issue by searching for the optimal control policy in a restricted parameterized policy space. We here focus on the case of discrete action space and introduce a novel policy parametrization that adopts particles to describe the map from the state space to the action space, each particle representing a region of the state space that is mapped into a certain action. The locations and actions associated with the particles describing a policy can be tuned by means of a recently introduced policy gradient method with parameter-based exploration. The task of selecting an appropriately sized set of particles is here solved through an iterative policy building scheme that adds new particles to improve the policy performance and is also capable of removing redundant particles. Experiments demonstrate the scalability of the proposed approach as the dimensionality of the state-space grows.

Most approaches for deadlock prevention and liveness enforcement in Petri nets rely on siphon control methods or the theory of regions to derive monitor-based supervisors. These techniques raise methodological and computational issues, from the existence of feasible solutions to the hardness of guaranteeing maximal permissivity and optimality in the size and cost of the control subnet. Recently, the supervisor design problem has also been reformulated as a direct monitor optimization task based on integer linear programming, which can more effectively deal with the mentioned issues and objectives. This paper introduces an efficient branch-and-bound scheme for the exploration of the solution space of the direct monitor optimization problem. An extensive computational analysis on a set of benchmark instances demonstrates the efficiency of the approach.

The characterization of complex air traffic situations is an important issue in air traffic management (ATM). Within the current ground-based ATM system, complexity metrics have been introduced with the goal of evaluating the difficulty experienced by air traffic controllers in guaranteeing the appropriate aircraft separation in a sector. The rapid increase in air travel demand calls for new generation ATM systems that can safely and efficiently handle higher levels of traffic. To this purpose, part of the responsibility for separation maintenance will be delegated to the aircraft, and trajectory management functions will be further automated and distributed. The evolution toward an autonomous aircraft framework envisages new tasks where assessing complexity may be valuable and requires a whole new perspective in the definition of suitable complexity metrics. This paper presents a critical analysis of the existing approaches for modeling and predicting air traffic complexity, examining their portability to autonomous ATM systems. Possible applications and related requirements will be discussed.

The extension of active noise control (ANC) techniques to deal with nonlinear effects such as distortion and saturation requires the introduction of suitable nonlinear model classes and adaptive algorithms. Large sized models are typically used, resulting in an increased computational load, delayed convergence (and sometimes even algorithm instability), and other unwanted dynamical effects due to overparametrization. This paper discusses the usage of polynomial nonlinear autoregressive models with exogenous variables (NARX) models and model selection techniques to reduce the model size and increase its robustness, for more efficient and reliable ANC. An offline procedure is devised to identify the controller model structure, and the controller parameters are successively updated with an adaptive algorithm based on the error gradient and on the residual noise. Simulation experiments show the effectiveness of the proposed approach. A brief analysis of the involved computational complexity is also provided.

In Petri-net (PN) modeling of flexible manufacturing systems, deadlock prevention is often addressed by means of siphon-control methods. Constraints that avoid the emptying of siphons can be easily implemented using additional places suitably connected to the PN transitions. Efficient siphon-based techniques achieve highly permissive solutions using as few control places as possible. One such technique employs a set-covering approach to optimally match emptiable siphons to critical markings. In this paper, a modified version of the method is proposed that achieves the same results in terms of permissivity and size of the control subnet but avoids full siphon enumeration. This greatly reduces the overall computational time and memory requirements and allows the applicability of the method to large-size models.

Deadlock prevention is a crucial step in the modeling of flexible manufacturing systems. In the Petri net framework, deadlock prevention policies based on siphon control are often employed, since it is easy to specify generalized mutual exclusion constraints that avoid the emptying of siphons. However, such policies may require an excessive computational load and result in impractical oversized control subnets. This is often a consequence of the redundancy in the control conditions derived from siphons. In this paper, a novel method is proposed that provides small size controllers, based on a set covering approach that conveniently relates siphons and markings. Some examples are provided to demonstrate the feasibility of the approach and to compare it with other methods proposed in the literature.

A crucial step in the logic-control design for flexible manufacturing systems is the modeling of the desired behavior in terms of production sequences. These are typically expressed as a series of informal specifications which are to be translated into a formal model. The present paper introduces a Petri net-based methodology for the modeling of production sequences conceived as an incremental process, which progressively adds subnets to a partial model until all specifications have been included. A formal verification of the model is performed at every stage, so that specifications, which are incongruent with the current model and can jeopardize its correctness, can be detected and rejected or modified. No unnecessary modeling constraints are imposed on the model class and behaviors characterized by complex alternative-parallel relationships between operations can be represented. A specific algorithm is provided for the automatic synthesis of complex connection structures in the model, allowing for greater flexibility and complexity in the design process. Note to practitioners - Petri-net (PN) models can be effectively used in the development and validation of industrial logic-control systems as long as well-structured modeling techniques are available that can be easily automated in a computer-aided design environment. This paper introduces a structured and modular novel design methodology based on PNs, which constructs the logic-control model in an incremental fashion, taking progressively into account the user's product-flow specifications, in terms of composition constraints on the operation execution. Each step of the design process increases the model according to rules which guarantee that it holds specific well-definiteness properties. In particular, the methodology prevents possible design mistakes from leading to deadlocks. From a practical point of view, the design process fits the step-by-step specification method that is commonly adopted quite well, with the great advantage of having immediate feedback on the formal correctness of the model at each step. Also, given a suitable implementation in software packages, the interaction with the user does not require specific knowledge of PNs. The actual control logic can then be obtained by adding the resource model to the product-flow model and detailing operations.

The paper addresses the problem of enumerating minimal siphons in an ordinary Petri net. The algorithms developed in this work recursively use a problem partitioning procedure to reduce the original search problem to multiple simpler search subproblems. Each subproblem has specific additional place constraints with respect to the original problem. Some results on algorithm correctness, convergence, and computational complexity are provided, as well as an experimental evaluation of performance. The algorithms can be applied to enumerate minimal, place-minimal siphons, or even siphons that are minimal with respect to given subsets of places.