The main scientific activities are focused on two main research field: Artificial Intelligence and the related problems concerning Knowledge Acquisition, Representation and Computation, and the area of Discrete Dynamical Systems, related to the modelling and the simulation of complex systems through Cellular Automata and Multi-agent Systems. In both the investigation area, the theoretical approach is joined to the experimental evaluation of the obtained results, and the application possibilities, in science and in the production reality. The main research interests are grouped in the following main points.

 Artificial Intelligence, Knowledge Representation and Management

Heterogeneous Knowledge Representation: integration of declarative and procedural knowledge through the introduction of the SA-Superposed Automata Networks (a class of Petri-nets) integrated to classical representation frameworks. The application fields to test the validity of the approach are medical diagnosis, and narrative knowledge. The main research effort in this area regards the development of innovative integrating approaches based on Bayes theory, fuzzy models and abduction applied to expert medical knowledge. Moreover, the creation of new generations of models to design and develop Second Generation Expert Systems has been deeply investigated and applied also in industrial environment. The development of many Knowledge-based Systems and Expert Systems (medicine, design, mechanics, business-services, environment and cultural heritage) allowed the identification of the notion of Knowledge Artifact to model and design knowledge-based solutions, and the adoption of ontology-based techniques to represent structural/functional knowledge integrated to heuristics (rule-based).Expert Systems

The integration of knowledge-based systems in complex software architectures has been studied in relation to the development of industrial projects for the monitoring and control of traffic on highways. The theoretical approach involved the problem of correlating alarm coming from distributed sensors through the use of computational models considering spatial and temporal reasoning (see below). Other research results have been obtained in the development of formal and computational models in mechanical design. Many industrial and research project allowed the test and evaluation of the developed approach to be performed.

Case Based Reasoning (CBR): the development of a model-based adapter module through substitutional adaptation techniques (ACM – Abstract Compound Model) in the CBR cycle is one of the main contributions in this research area. It has been developed within the P-Race Project (funded by Pirelli Tires) dedicated to the development of an innovative computer-based system for the chemical formulation of rubber compounds for motorsports tires. In the second project developed for Pirelli Tires (P-Truck) an original approach in the ontological approach in cases description has been introduced and integrated to the chemical formulation module. Major research results within this area concern the introduction of fuzzy indexing and retrieval on the cases-memory. Moreover, the investigation of the CBR approach in Core Knowledge Management to support the design of innovative products has been developed and applied to the referred Community of Practice. The development of the original conceptual and computational framework based on Complex Knowledge Structures (CKS) allowed the development of new research approaches in the integration of knowledge acquisition methodologies and knowledge representation tools.

Spatial-temporal Knowledge Representation: this research area regards the development of conceptual, formal and computational frameworks to model and compute spatial-temporal techniques (S-T Logic) to support the design and the application of complex control and monitoring systems. The main results concern the creation of a formal abductive-reasoning framework to support the correlation of interpreted data coming from distributed sensors in the field of context-aware pervasive and ubiquitous computing. Application results (exploited by Project Automation) are actually used for the traffic control through 4 installations on the Italian Highway system.


Discrete Dynamical Simulation, Cellular Automata, Multi-agent Systems

Cellular Automata: this research area has been investigated both on the theoretical and application sides. Theoretical results within this fields concern the definition of multilayered automata networks tested on biological and physical systems: reaction-diffusion systems, intracellular dynamics of the calcium ion, cellular and humoral interaction in the immune system, vegetal and animal population dynamics; radiation-matter interaction, visco-elastic properties of chemical compounds, percolation processes in porous media. Other theoretical results concern the development of a neuron-genetic framework for pattern recognition, and for the study of the dynamical evolution of transition states rules in Cellular Automata.

Multi-agent Systems: the main contributions of the research within this area concern the development of an original modeling and computational Agent-based approach integrating the synchronous capabilities of Cellular Automata and at-a-distance interaction mechanisms (Situated Cellular Agents  - SCA). One of the main characteristics of this model concern the explicit representation of spatial features within the computational environment. Theoretical results in this area allowed testing the development of formal, representational and computational tools on many cases of complex dynamical discrete systems, in order to study (and analyze) emergent properties occurring during the dynamical behavior of interacting agents (vegetal populations dynamics, cooperative software applications, cellular interaction in the immune system). The main research effort within this research area has been focused in the study of the dynamical properties of pedestrian and crowds. The SCA approach has been successfully applied in the modeling and simulation of indoor and outdoor crowd phenomena, in order to study paradigmatic cases of emergent coordination in interaction conditions. Moreover, methodological and multidisciplinary approaches (integrated with studies coming from psychology and sociology) for data acquisition on crowd have been studied in order to handle real cases: sports, concerts, shopping centers, airports).

The SCA-based model has been tested to study emergent dynamical features in ubiquitous computing, and adaptive software architectures. The experimental activity is conducted on specific software platform implementing the developed Agent-based approach, both in 2D and 3D computational environment.

Development of a novel Agent-based computational model for the simulation of complex behaviors of groups and pedestrian based on proxemic distances, and application in the case of the new Mashaer trail line in Arafat (Saudi Arabia) during the Hajj.

The creation of Agent-based computational models for pedestrians and groups has been focused also for cases of mobility and transportation in the Ageing Society, in order to face the development of integrated software platforms allowing the design of facilitation for aged people.

 

© Stefania Bandini 2011