Paulo Urbano | Universidade de Lisboa (original) (raw)

Papers by Paulo Urbano

Mathematical and Computational Applications

Robotics technology has made significant advancements in various fields in industry and society. ... more Robotics technology has made significant advancements in various fields in industry and society. It is clear how robotics has transformed manufacturing processes and increased productivity. Additionally, navigation robotics has also been impacted by these advancements, with investors now investing in autonomous transportation for both public and private use. This research aims to explore how training scenarios affect the learning process for autonomous navigation tasks. The primary objective is to address whether the initial conditions (learning cases) have a positive or negative impact on the ability to develop general controllers. By examining this research question, the study seeks to provide insights into how to optimize the training process for autonomous navigation tasks, ultimately improving the quality of the controllers that are developed. Through this investigation, the study aims to contribute to the broader goal of advancing the field of autonomous navigation and develop...

Lecture Notes in Computer Science, 2018

In evolutionary robotics role allocation studies, it is common that the role assumed by each robo... more In evolutionary robotics role allocation studies, it is common that the role assumed by each robot is strongly associated with specific local conditions, which may compromise scalability and robustness because of the dependency on those conditions. To increase scalability, communication has been proposed as a means for robots to exchange signals that represent roles. This idea was successfully applied to evolve communication-based role allocation for a two-role task. However, it was necessary to reward signal differentiation in the fitness function, which is a serious limitation as it does not generalize to tasks where the number of roles is unknown a priori. In this paper, we show that rewarding signal differentiation is not necessary to evolve communication-based role allocation strategies for the given task, and we improve reported scalability, while requiring less a priori knowledge. Our approach puts fewer constrains on the evolutionary process and enhances the potential of evolving communication-based role allocation for more complex tasks.

Lecture Notes in Computer Science, 2015

Maritime domain awareness deals with the situational understanding of maritime activities that co... more Maritime domain awareness deals with the situational understanding of maritime activities that could impact the security, safety, economy or environment. It enables quick threat identification, informed decision making, effective action support, knowledge sharing and more accurate situational awareness. In this paper, we propose a novel computational intelligence framework, based on genetic programming, to predict the position of vessels, based on information related to the vessels past positions in a specific time interval. Given the complexity of the task, two well known improvements of genetic programming, namely geometric semantic operators and linear scaling, are integrated in a new and sophisticated genetic programming system. The work has many objectives, for instance assisting more quickly and effectively a vessel when an emergency arises or being able to chase more efficiently a vessel that is accomplishing illegal actions. The proposed system has been compared to two different versions of genetic programming and three non-evolutionary machine learning methods, outperforming all of them on all the studied test cases.

In this paper we present De.:SID, a rule-based Intelligent Online Survey program. We have incorpo... more In this paper we present De.:SID, a rule-based Intelligent Online Survey program. We have incorporated three rule knowledge bases in a standard online survey architecture that are used to control three vital components on a survey: (1) the dependencies between questions, i.e. the structure and survey branching logic, (2) the decision regarding the selection of the next question to be asked, and (3) the inconsistency detection between answers to different questions. These rule-based components allow us to escape a predetermined question sequence, achieving flexibility and adaptability to the user’s answers; besides, they enhance usability allowing an easy navigation along the different survey questions and the possibility to backtrack and revise the answers, at any moment, without loosing global coherence. There is an explicit treatment of inconsistent situations by exposing them and inviting the user to revise his answers. De.:SID benefits from the qualities that are generally assoc...

Computational Generative Art has been inspired by complex collective tasks made by social insects... more Computational Generative Art has been inspired by complex collective tasks made by social insects like the ants, which are able to coordinate through local interactions and simple stochastic behavior. In this paper we present the Light Show, an application of the mechanism of flash synchronization exhibited by some species of fireflies. The virtual fireflies from The Light Show gather and fly over digital readymades, self-choreographing the rhythm of illumination of their artistic habitats. We present a standard model with some design parameters able to control synchronization and also a variation able to exhibit clusters of sync at different phases that grow, fight, disappear or win, illuminating different parts of a digital image in an animated process.

Soft Computing, 2016

Over recent years, evolutionary computation research has begun to emphasize the issue of generali... more Over recent years, evolutionary computation research has begun to emphasize the issue of generalization. Instead of evolving solutions that are optimized for a particular problem instance, the goal is to evolve solutions that can generalize to various different scenarios. This paper compares objective-based search and novelty search on a set of generalization oriented experiments for a navigation task using grammatical evolution (GE). In particular, this paper studies the impact that the training set has on the generalization of evolved solutions, considering: (1) the training set size; (2) the manner in which the training set is chosen (random or manual); and (3) if the training set is fixed throughout the run or dynamically changed every generation. Experimental results suggest that novelty search outperforms objective-based search in terms of evolving navigation behaviors that are able to cope with different initial conditions. The traditional objective-based search requires larger training sets and its performance degrades when the training set is not fixed. On the other hand, novelty search seems to be robust to different training sets, finding general solutions in almost all of the studied conditions with almost perfect generalization in many scenarios.

Leonardo

Soma 416 frame 159, screen image from volatile sequence, 4200 × 600 pixels, 2005. (© Tim Barrass)... more Soma 416 frame 159, screen image from volatile sequence, 4200 × 600 pixels, 2005. (© Tim Barrass) Soma 435 frame 026, screen image from volatile sequence, 4200 × 600 pixels, 2005. (© Tim Barrass) SOMA is a software model of a dynamic system in which a virtual population of ant-like drawing agents develops individual behaviors in response to marks on a surface that they collectively modify. A small neural network called a Kohonen self-organizing map (SOM), which responds to nearby patterns on the drawing surface, regulates the motion of each ant. The SOM determines the ant’s next move and is modified by the most recent pattern in the process. Each ant leaves a trail, contributing to the overall image. When thousands of ants interact this way, a complex multi-directional feedback system in which agents indirectly influence one another’s internal structures forms through the effects the ants have on their surroundings. It is difficult to predict what will happen without running the sys...

Lecture Notes in Computer Science, 2014

Recent research on evolutionary algorithms has begun to focus on the issue of generalization. Whi... more Recent research on evolutionary algorithms has begun to focus on the issue of generalization. While most works emphasize the evolution of high quality solutions for particular problem instances, others are addressing the issue of evolving solutions that can generalize in different scenarios, which is also the focus of the present paper. In particular, this paper compares fitness-based search, Novelty Search (NS), and random search in a set of generalization oriented experiments in a maze navigation problem using Grammatical Evolution (GE), a variant of Genetic Programming. Experimental results suggest that NS outperforms the other search methods in terms of evolving general navigation behaviors that are able to cope with different initial conditions within a static deceptive maze.

Abstract. We propose and evaluate a novel approach to the on-line synthesis of neural controllers... more Abstract. We propose and evaluate a novel approach to the on-line synthesis of neural controllers for groups and swarms of au-tonomous robots. We combine online evolution of weights and net-work topology with neuromodulated learning in a completely de-centralised manner. We demonstrate our method through a series of simulation-based experiments in which a group of e-puck-like robots must perform a dynamic concurrent foraging task. In this task, scat-tered food items periodically change their nutritive value or become poisonous. Our results show that when neuromodulated learning is employed, neural controllers are synthesised faster than by evolution alone. We demonstrate that the online evolutionary process is capable of generating controllers well adapted to the periodic task changes. We evaluate the performance both in a single robot setup and in a multirobot setup. An analysis of the evolved networks shows that they are characterised by specialised modulatory neurons that exclu-sively regulate online learning in the output neurons. 1

Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), 2012

ABSTRACT In this paper we introduce the concept of a progressive percussion graph as a musical sp... more ABSTRACT In this paper we introduce the concept of a progressive percussion graph as a musical space and the metaphor of composition as the musical expression of a traveling experience in that space. A Progressive Percussion Graph is a directed graph where each node is associated with a particular percussion rhythm and each connection corresponds to a rhythmic progression, generated through optimization processes, from one percussion rhythm to another, respecting the connections direction. We have explored different optimization techniques and different path-finding algorithms resulting in a rich and diverse musical output.

We present 4CitySemantics, a computational tool for City Planning. The main objective of this too... more We present 4CitySemantics, a computational tool for City Planning. The main objective of this tool is to assist the participants of the urban development process in constructing adequate interpretations of a defined intervention site and its surrounding area (buffer), thereby contributing for better planning. The function of 4CitySemantics is to identify and classify both geographically and semantically the urban areas subjected to the intervention, which can be visualized. The key principle behind the tool is to offer the user as much flexibility as possible, so that s/he can develop customized semantic interpretations of the intervention areas in order to formulate an adequate intervention program. Such flexibility is achieved through the ample use of customizable ontologies for interpreting population and land use data, separating urban knowledge from the application tool, which can easily adapt to different urban semantic standards. By using the functionalities of the tool backe...

The Exquisite Corpse is a game-based art form popularized by the surrealists. The final result is... more The Exquisite Corpse is a game-based art form popularized by the surrealists. The final result is based on an unconscious collaboration of collective artists; each provides his or her part without knowing what the other has selected. For example, each one provides adjective, verb and noun to create poems; head, torso and legs for drawings. Actually, concerning drawings, each one draws his part in a piece of paper and folds it, so the next artist is able to see the border traces in the folded column and normally begins from there. The exquisite Corpse is a game with several interesting properties: 1. It’s an example of collective creation and formation of a collective pattern; 2. We can envision a micro and macro levels, even if there are not so many participants 3. There is no unity, there is author fragmentation and so the complete outcome cannot be intended and planned in advance; 4. In fact, we are in face of an unpredictable and surprising outcome; 5. There is no direct communic...

Proceedings of the 2020 Genetic and Evolutionary Computation Conference Companion, 2020

In this work, we analyse the performance of Novelty Search (NS) in a set of generalization experi... more In this work, we analyse the performance of Novelty Search (NS) in a set of generalization experiments in a navigation task with Grammatical Evolution. Agents are trained on a single, simple environment, and tested on a selection of related, increasingly more difficult environments. We show that agents discovered with NS, although using a tiny number (six) of training samples, successfully generalise to these more difficult environments.

Proceedings of the 3rd International Conference on Agents and Artificial Intelligence, 2011

Social conventions are useful for the coordination of multi-agent systems. Decentralized models o... more Social conventions are useful for the coordination of multi-agent systems. Decentralized models of social convention emergence have demonstrated that global agreement can be the result of local coordination behaviors without the need for any central control and authority. Convention arises through a co-learning process from repeated interactions, where the history of interactions plays a fundamental role in the learning process. The main research goal of this work is to study the role of ties in the standard frequency model called External Majority (EM). In the External Majority case agents change to a new convention only if a different convention was more often seen than the current one in the last μ interactions. Agents prefer to conserve their conventions if the current one is included in the set of the most often seen in the last μ encounters. We study three variations in EM behaviors regarding the way of dealing with tie situations and study empirically their impact on convention emergence efficiency. Efficiency is a decisive property in what concerns the design of large-scale self-organizing artificial systems, and one of the variations we propose strongly improves consensus emergence performance.

Inteligencia Artificial, 2020

In evolutionary robotics role allocation studies, it is common that the role assumed by each robo... more In evolutionary robotics role allocation studies, it is common that the role assumed by each robot is strongly associated with specific local conditions, which may compromise scalability and robustness because of the dependency on those conditions. To increase scalability, communication has been proposed as a means for robots to exchange signals that represent roles. This idea was successfully applied to evolve communication-based role allocation for a two-role task. However, it was necessary to reward signal differentiation in the fitness function, which is a serious limitation as it does not generalize to tasks where the number of roles is unknown a priori. In this paper, we show that rewarding signal differentiation is not necessary to evolve communication-based role allocation strategies for the given task, and we improve reported scalability, while requiring less a priori knowledge. Our approach for the two-role task puts fewer constrains on the evolutionary process and enhance...

Mathematical and Computational Applications

Lecture Notes in Computer Science, 2018

Lecture Notes in Computer Science, 2015

Soft Computing, 2016

Leonardo

Lecture Notes in Computer Science, 2014

Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), 2012

Proceedings of the 2020 Genetic and Evolutionary Computation Conference Companion, 2020

Proceedings of the 3rd International Conference on Agents and Artificial Intelligence, 2011

Inteligencia Artificial, 2020

In evolutionary robotics role allocation studies, it is common that the role assumed by each robo... more In evolutionary robotics role allocation studies, it is common that the role assumed by each robot is strongly associated with specific local conditions, which may compromise scalability and robustness because of the dependency on those conditions. To increase scalability, communication has been proposed as a means for robots to exchange signals that represent roles. This idea was successfully applied to evolve communication-based role allocation for a two-role task. However, it was necessary to reward signal differentiation in the fitness function, which is a serious limitation as it does not generalize to tasks where the number of roles is unknown a priori. In this paper, we show that rewarding signal differentiation is not necessary to evolve communication-based role allocation strategies for the given task, and we improve reported scalability, while requiring less a priori knowledge. Our approach for the two-role task puts fewer constrains on the evolutionary process and enhance...