Author name cluster

Angelo Ferrando 0001

Possible papers associated with this exact author name in Arrow. This page groups case-insensitive exact name matches and is not a full identity disambiguation profile.

10 papers

1 author row

ECAI Conference 2025 Conference Paper

Let Me Talk to You! Natural Language Interaction Between Humans and BDI Agents via ChatBDI

Andrea Gatti 0002
Viviana Mascardi
Angelo Ferrando 0001

In this paper we describe ChatBDI, a framework for extending Belief-Desire-Intention (BDI) agents implemented in Jason with the ability to understand and generate messages in natural language by exploiting embeddings and Large Language Models (LLMs). Thanks to the generative power of LLMs, the ‘chattification’ of new or legacy multiagent systems (MAS) adds a creative and fluent ‘language actuator’ to BDI agents and serves two main purposes. First, it allows users to enter the MAS and conversate with any other software agent in natural language, as if humans were agents themselves. Second, it allows both the MAS developers and the users to follow the conversation among software agents and to ask them information on their behavior and decisions, acting as a lightweight co-pilot and improving transparency and explainability. The major strength of ChatBDI, and its distinguishing feature w. r. t. related works, is its general purpose nature: by exploiting plan injection and sophisticated meta-programming facilities, ChatBDI can be used to chattify any MAS implemented in Jason or JaCaMo with neither adaptations of the ChatBDI code itself, nor changes to the existing AgentSpeak(L) agents’ source code.

Details

ECAI Conference 2025 Conference Paper

Reliable Intention Selection in BDI Agents with Recovery Shields

Angelo Ferrando 0001
Rafael C. Cardoso 0001

The existing approaches to enforcing runtime properties and handling failures in autonomous agents primarily focus on single-agent systems, responding to violations by immediately rejecting unsafe actions. In this paper, we extend the notion of safety shields for Belief-Desire-Intention agents by proposing a revised version where a shield can observe and respond not only to what the agent itself does, but also to changes from other agents, when these affect the shielded agent’s behaviour. Our shields suspend intentions that would break a formal specification and resume them once it is safe to do so. To support this, we introduce recovery shields, a new mechanism that defines when a suspended intention can be safely resumed. Our main contributions are extensions to the reasoning cycle and operational semantics of AgentSpeak(L), as well as an implementation in the JaCaMo platform.

Details

ECAI Conference 2025 Conference Paper

Runtime Verification with Rational Multi-Monitors

Davide Catta
Angelo Ferrando 0001
Vadim Malvone

Runtime verification (RV) is a lightweight technique for checking system correctness against formal specifications. Traditional RV assumes full system observability, which rarely holds in distributed and component-based systems where monitors only see partial traces. This leads to inconclusive or incorrect verdicts. To address this, recent work has explored monitors that handle imperfect information and reason about visibility. However, these approaches focus on isolated monitors and overlook coordination in distributed settings. We propose a novel framework for runtime verification with rational multi-monitors, where each monitor is a resource-bounded agent with a local specification. Monitors strategically decide what information to share or request, balancing verification goals with communication costs. We formalise this interaction using multi-agent system techniques and synthesise cooperative strategies through model checking. We implement our approach and evaluate it in a case study, showing that rational coordination improves monitoring conclusiveness over existing approaches.

Details

EUMAS Conference 2024 Conference Paper

MAiS: Exploiting JADE as a Multi-agent Simulator of the Immune System

Sanchayan Bhunia
Angelo Ferrando 0001
Viviana Mascardi
Chiara Vitale

Abstract The immune system is the second most complex biological system after the brain. It consists in millions of cells, of various nature, interacting amongst them to keep the organism safe from external enemies (pathogens), such as viruses and bacteria. To better understand how the immune system works, and how it reacts to certain diseases and cures, simulations have been proposed over the years. Amongst them, we may find agent-based ones where the organism’s actors, like cells, antibodies, viruses, and so on, are represented as agents. In this paper, we present the initial design and development of an agent-based simulation of the immune system using a well-known agent framework, JADE. We present the engineering choices we made and the instantiation of some steps of the secondary immune system response. We discuss the implementation in JADE, and we present some experimental results.

Details

EUMAS Conference 2023 Conference Paper

AGAMAS: A New Agent-Oriented Traffic Simulation Framework for SUMO

Mahyar Sadeghi Garjan
Tommy Chaanine
Cecilia Pasquale
Vito Paolo Pastore
Angelo Ferrando 0001

Abstract Simulating everyday traffic scenarios is not an easy task. Many aspects have to be taken into consideration and properly modelled, from static components, like traffic lights, to dynamic components, like vehicles. Due to their intrinsic autonomy and distribution, such components have already been designed as software agents, and integrated into existing traffic simulators, such as SUMO. The needing for agent-based modelling is even more evident when autonomous vehicles are present in the simulation. In this paper, we present an Agent-Based Traffic Simulation framework, where the simulation components can be defined as JADE agents. We present the engineering of our framework, and we show how it represents a new alternative for creating Agent-Based simulations in the largely used SUMO traffic simulator. We also demonstrate its applicability by employing the framework in one case study involving autonomous vehicles.

Details

ECAI Conference 2023 Conference Paper

Failure Handling in BDI Plans via Runtime Enforcement

Angelo Ferrando 0001
Rafael C. Cardoso 0001

Engineering a software system can be a complex process and prone to failure. This is exacerbated when the system under consideration presents some degree of autonomy, such as in cognitive agents. In this paper, we use runtime verification as a way to enforce safety properties on Belief-Desire-Intention (BDI) agents by enveloping certain plans in safety shields. These shields function as a failure handling mechanism, they can detect and avoid violations in shielded plans. The safety shields also provide automated failure recovery by attempting alternative execution paths to avoid violations.

Details

EUMAS Conference 2023 Conference Paper

Integrating Ontologies and Cognitive Conversational Agents in On2Conv

Zeinab Namakizadeh Esfahani
Débora Cristina Engelmann
Angelo Ferrando 0001
Massimiliano Margarone
Viviana Mascardi

Abstract Multiagent systems have been successfully used in many domains. Being social, they are expected to communicate with human users in natural language. Nevertheless, the natural interaction between agents and humans is still challenging. Chatbot technologies are a key enabler to boost the communication between humans and software agents, but few technical solutions exist that make the agents’ reasoning capabilities easily accessible by a human user via a chatbot and, on the other hand, the chatbot’s answers more controllable and explainable. Dial4JaCa is one of such tools. It creates a bridge between Dialogflow and the JaCaMo cognitive-oriented and symbolic AI-based framework: the user’s interface is a Dialogflow chatbot allowing the user to interact in natural language, and the backend implementing the reasoning and performing required actions is a JaCaMo agent. However, in Dial4JaCa the consistency between data that feed the JaCaMo agent and those that feed the Dialogflow chatbot must be guaranteed by the developer via an error-prone and tedious manual process. By taking an ontology describing the domain of interest in input and generating both the skeleton for the JaCaMo agent’s behaviour and the intents for the Dialogflow chatbot, On2Conv improves Dial4JaCa robustness and reliability, and moves one step towards an explainable integration of agents and chatbots.

Details

EUMAS Conference 2023 Conference Paper

Using a BDI Agent to Represent a Human on the Factory Floor of the ARIAC 2023 Industrial Automation Competition

Leandro Buss Becker
Anthony Downs
Craig Schlenoff
Justin Albrecht
Zeid Kootbally
Angelo Ferrando 0001
Rafael C. Cardoso 0001
Michael Fisher 0001

Abstract The “Agile Robotics for Industrial Automation Competition” (ARIAC) is an international robotic competition carried out in a simulated factory floor using ROS 2 (Robot Operating System)/Gazebo. Competitors control one gantry robot, four AGVs, and many other elements/devices, overcoming a range of agility challenges in this simulated environment, and are provided with a scoring system to evaluate their performance during the tasks. This paper describes one of the agility challenges in ARIAC 2023, which pertains to a simulated human operator on the factory floor. In undertaking manufacturing tasks, competitors must avoid close proximity between the gantry robot and the human not to get penalized. The human operator is implemented as a Belief-Desire-Intention (BDI) agent in Jason. It is provided with a range of different potential types of behaviour in what concerns with how such human reacts when in proximity to the gantry robot. Three different personalities are presented, ranging from a minimally intrusive up to a very intrusive one. A preliminary analysis was conducted to evaluate the impact of using the developed Jason agent in the ARIAC 2023 competition.

Details

EUMAS Conference 2022 Conference Paper

Mind the Gap! Runtime Verification of Partially Observable MASs with Probabilistic Trace Expressions

Davide Ancona
Angelo Ferrando 0001
Viviana Mascardi

Abstract In this paper we present the theory behind Probabilistic Trace Expressions (PTEs), an extension of Trace Expressions where types of events that can be observed by a monitor are associated with an observation probability. PTEs can be exploited for monitoring that agents in a MAS interact in compliance with an Agent Interaction Protocol (AIP) modeled as a PTE, even when the monitor realizes that an interaction took place in the MAS, but it was not correctly observed (“observation gap”). To this aim, we adapt an existing approach for runtime verification with state estimation, we present a semantics for PTEs that allows for the estimation of the probability to reach a given state, given a sequence of observations which may include observation gaps, we present a centralized implemented algorithm to dynamically verify the behavior of the MAS under monitoring and we discuss its potential and limitations.

Details

EUMAS Conference 2020 Conference Paper

An Interface for Programming Verifiable Autonomous Agents in ROS

Rafael C. Cardoso 0001
Angelo Ferrando 0001
Louise A. Dennis
Michael Fisher 0001

Abstract Autonomy has been one of the most desirable features for robotic applications in recent years. This is evidenced by a recent surge of research in autonomous driving cars, strong government funding for research in robotics for extreme environments, and overall progress in service robots. Autonomous decision-making is often at the core of these systems, thus, it is important to be able to verify and validate properties that relate to the correct behaviour that is expected of the system. Our main contribution in this paper, is an interface for integrating BDI-based agents into robotic systems developed using ROS. We use the G wendolen language to program our BDI agents and to make use of the AJPF model checker in order to verify properties related to the decision-making in the agent programs. Our case studies include 3D simulations using a simple autonomous patrolling behaviour of a TurtleBot, and multiple TurtleBots servicing a house that can cooperate with each other in case of failure.

Details DOI