User Tools

Site Tools


Research Interests

Generally speaking the areas of my interest include modelling and simulation of complex systems and stochastic processes, systems optimisation, Distributed Artificial Intelligence (DAI), scientific visualisation, engineering software and CAD. I am currently focusing my research on the application of DAI techniques to tackle engineering problems, especially using multi-agent systems as a modelling metaphor to address issues in Artificial Transportation Systems. More specifically, I am involved with projects related to:

  • Multi-paradigm traffic simulation;
  • Pedestrian simulation;
  • Behavioural modelling and social simulation;
  • Validation and calibration methodologies for MAS-based simulation;
  • Spatio-temporal data mining, analysis and visualisation;
  • GIS-T and GIS-based simulation;
  • UbiComp, Pervasive Systems and Ambient Intelligence applied to mobility systems.

Some On-going Research Projects

Below you can find a brief description of on-going projects. Demos are available on YouTube. For any further information, please to not hesitate to contact me.

ModP: a Pedestrian Modelling Platform

The ModP prototype is the first extension to the MAS-Ter Lab platform with the intent of including aspects related to pedestrian simulation into the framework, which was originally thought to simulate vehicular flow solely. The aim of such an extension is two-fold. First, we intend to integrate pedestrian behaviour with traffic interactions at a microscopic level at the urban network. Second, we intend to study people interactions in other diverse scenarios, not only within traffic networks, namely in situations such as evacuations from public sites, to better plan public spaces, as well as a decision-support system to improve CAD (computer-aided design) tools used in most building project software. This first prototype resulted from collaborations with TRENMON Eng. (an industrial partner), and soon after captivated the interested of the National Laboratory of Civil Engineering – LNEC.

Mod-P: pedestrian simulation using autonomous agents

SPEED: Simulation of Pedestrians and Elicitation of their Emergent Dynamics

The Speed framework aims to study pedestrian dynamics and interactions, which concern reasoning processes, (path) planning, and all other aspects associated with the pedestrian movement in a variety of mobility settings, both indoors and outdoors. This kind of tool is important for urban planners as an aid for designing and evaluating urban spaces regarding comfort, safety, and other important issues, such as accessibility of public buildings. The behavior elicitation part integrates agent-based modeling, social simulation, and serious games, which is used both as a training tool and an important observation aid. PDA-based agents capture players’ behavior during the game and are used to form a synthetic population upon which social simulations reproducing various situations and what-if scenarios can be performed. The serious game component is implemented in Unity3D, whereas the agent-based social simulation may be implemented either in ModP or in NetLogo.

SPEED: EVA instance for evacuation behaviour analysis

TrasMAPI: an Agent-based API to support multi-paradigm traffic simulation

The TrasMAPI framework is a Java API whose main objective is to support the real-time integration of multiple traffic simulators, as well as the implementation of agent-based solutions as a means to better control and mange traffic scenarios in different road network scenarios. A first version of the API is ready and support multi-scenario simulation using the open-source simulator SUMO. In its current version, it is possible to implement autonomous agent controllers that govern signal controller intersections, allowing for the implementation of different AI-based techniques to improve control strategies and coordination policies. Other simulators are to be supported, such as Leeds University’s DRACULA microscopic simulator, as well as commercial ones such as Paramics and AimSun.

SUMO and JADE integration through TraSMAPI

Source code and manuals available through FEUP VPN

IC-DEEP: In-Car Driver Ergonomics Evaluation Platform

In-vehicle information systems and advanced driver assistance systems are used during the context of driving, where the objective is to deliver a better driving experience by providing infotainment functionalities and enhancing safety, respectively. As the inherent complexity of such systems in the safety-critical context of driving increases, the role of ergonomics and human-computer interactions must be carefully analyzed and understood. The IC-DEEP framework was created to meet the demand for rapid design-prototype-evaluate cycles in the development of in-vehicle information systems and advanced driver assistance systems, as well as to improve driver modeling in microscopic simulation. This low-cost platform is composed of commodity hardware, game accessories, and a serious game (Serious Driving) developed in Unity3D. The PDA concept is used to collect driver-performance statistics in both primary and secondary tasks and elicit their behavior during the game, allowing the generation of a synthetic population of PDA-based driver agents whose interactions are simulated in an agent-based microscopic simulator for further analysis and other studies with different purposes.

Use of IC-DEEP for the analysis for safety distance between vehicles

MAS-Ter VD Core

The MAS-Ter Lab VD Core is a multi-agent microscopic traffic simulator that underlies the whole structure of the MAS-Ter Lab Virtual Domain subsystem. Drivers are represented as agents, as well as are control systems, such as semaphores at traffic junctions. The simulator allows different traffic control strategies to be implemented and attached to the simulator through sockets-based integration. The enhanced architecture of MAS-Ter Lab VD Core follows the IEEE HLA Standards for distributed simulation and integrates JADE, for the implementation of a multi-agent population, and SUMO, for the microscopic simulation of vehicle interactions.

HLA-based implementation of VD Core using SUMO as microscopic simulator

NetEditor: the MAS-Ter Lab generic network editor

The NetEditor is a collaborative network editor that allows urban traffic networks to be modelled and exported into different microscopic traffic simulators, including the MAS-Ter Lab VD Core microscopic simulator and the open-source SUMO simulator. In addition, the architecture of the NetEditor allows multiple users, from different disciplines within Transportation Engineering, to interact over a common representation of the application domain, supported by a GIS database.

NetEditor - generic traffic network editor

Student Supervision and Tutoring

PhD Students

  • Lúcio Sanchez Passos (ProDEI, co-supervision with J. Gabriel)
  • Zafeiris Kokkinogenis (ProDEI, co-supervision with J. Gabriel and M. Ferreira)
  • João Almeida (ProDEI, co-supervision with A. Leça)
  • José Luis Ferras Pereira (ProDEI, co-supervision with J. Barros)
  • Deborah Perrotta de Andrade (DLIT, co-supervision with J. Afonso)

MSc Students

  • Daniel Almeida (MIEIC, co-supervision with L. Sarmento)
  • Marco Costa (MIEIC, co-supervision with A. Coelho)
  • João Magalhães (MIEEC, co-supervision with P. Fortuna)
  • José Luis Pereira (MIEEC)
  • Marco Paiva (MIEEC, co-supervision with M. Petry)
  • Jilson Moreira (MIEEC)

Former Students

Prospective Students

If you are interested in doing research on topics related to my research activities, I will be glad to discuss with you about options and opportunities for pursuing an MSc or a PhD under my supervision. Post-doc projects can also be discussed and are mostly welcome. Scholarships are not always available and will very much depend on the merit of each candidate. FCT usually issues two calls for PhD grants each year, whereas post-doc grant proposals can usually be submitted on a continuous basis. Please, refer to FCT's Web site for more information on eligibility. If you are a foreign student, please check and look for support from your own country first. Some international students have joined our group under the ERASMUS and the IAESTE programmes, for instance.

research/start.txt · Last modified: 2015/08/25 19:46 by Rosaldo J. F. Rossetti