Oscar J. Romero
PhD., Project Scientist
oscarr@andrew.cmu.edu
I am fascinated about the idea of building intelligent systems that can help humans (and why not, other intelligent agents) on their daily tasks. This goal is ambitious, challenging, inspiring, and best of all, multi-disciplinary: it lies at a pivotal intersection between areas that I am really interested in, such as AI, cognitive science, psychology, biology, neurophysiology, linguistics, adaptive behavior, computer science, and software engineering.
My research interests mainly focus on topics related to Artificial Cognition & Cognitive Architectures, Machine Learning, Computational Intelligence, Hybrid Systems, Evolutionary Computation, Intelligent Autonomous Agents (e.g., personal assistants, robotic systems, intelligent tutoring systems, conversational agents, multi agent systems, etc.), and Large-scale System Design. My goal is to build robust and complex software systems that integrate all of these areas in a synergistic manner in order to solve real-world problems.
I have a solid foundation in computer science, with strong competencies in data structures, algorithms, cognitive modeling, and software engineering. During the last 20 years of my career I have participated in recognized projects on both research and software industry. I have performed roles ranging from Developer to Research Project Scientist, led teams of up to 30 people, managed projects following maturity models and quality standards, and defined critical architectural aspects of complex software projects.
The projects that I have worked and conducted research on can be categorized as follows:
- Intelligent Personal Assistants
- Conversational Systems
- Tutoring/Pedagogical Multi-Agent Systems
- Cognitive Architectures for Autonom. Agents
- Dynamic Service Composition
- Cognitive Bio-Robotics
- Social Behavior Modeling
- Semantic Web and Intelligent Tagging
- Artificial Life
Education.
- Major in: AI, Computer Science, and Sw. Engineering
- Minor in: Cybernetics, Project Management, and Systems Theory.
- Final Project: Intelligent Tutoring Systems for Arithmetic
- Major in: AI, Agents and Multi-Agent Systems, and Software Engineering
- Minor in: Virtual Intelligent Systems and Artificial Life.
- Final Project: Animats (Artificial Animals) using bio-inspired computational intelligence
- Major in: AI, Artificial Cognition, and Software Engineering.
- Minor in: Agents and Multi-Agent Systems, and Robotic Systems.
- Final Project: Cognitive Architecture for Intelligent Agents (Summa Cum Laude)
Work.
Project Scientist
Details:
- Institution: Carnegie Mellon University (USA)
- Department: Robotics Institute
- Lab: TBD (Transportation, Bots and Disability)
- Research: Conversational Systems, Neuro-symbolic Architectures
- Projects: Accessible Augmented-Reality Conversational System
Overview:
Architecting and building an adaptive conversational system to support people with/without disabilities that use transit information apps, by embracing universal design. I advance research on hybrid neuro-symbolic approaches for conversational design by merging both transformer neural language models and symbolic reasoning.
Year: 2019 - Present
Research Project Scientist
Details:
- Institution: Carnegie Mellon University (USA)
- Department: Machine Learning
- Research: Intelligent Agents for Mobile, Machine Learning, Behavior Orchestration, Cognitive Architectures
- Projects: InMind, SARA
Overview:
My research work focuses on developing a software architecture for an intelligent agent (called InMind) that resides inside a mobile device (e.g., smartphone), that serves as an intelligent assistant to its user while automatically learning his/her interests, habits, and preferences. I am responsible of defining a multi-agent cognitive architecture that allows agents to interact in a pervasive computing environment, make decisions, create plans that orchestrate dynamic service composition, filter information through an attenctional mechanims, and learn from environmental regularities.
Year: 2014 - 2019
Research Associate
Details:
- Institution: Carnegie Mellon University (USA)
- Department: Psychology
- Activities: Development of Software Components, Generation of Technical Reports, Writing scientific papers
- Projects: RCTA, Cognitve Modeling
Overview:
As a Research Associate in the Psychology Dept. at CMU, I developed computational cognitive models of intelligence using the ACT-R cognitive architecture. I have modeled a diverse range of tasks, including behavior in social networks and mission planning in robotic settings. I have conducted research on topics such as: Artificial Cognition and Cognitive Modeling, Intelligent Multi-agent Systems, Complex Behavior, Robotics, Machine Learning, Decision Making, Automated Mission Planning, Network Science, and Social Behavior Simulation.
Year: 2013 - 2014
Research Associate - Postdoc
Details:
- Institution: Technical University of Madrid (Spain)
- Department: Computer Science
- Projects: MILES, CHIL2, REHABILITA
Overview:
As a Research Associate in the Computer Science Dept., I conducted research activities such as: developing a multi-agent platform for cognitive agents, implementing systems based on ontology engineering and intelligent tagging, development of knowledge-based personal assistants for medical rehabilitation over mobile apps, development of intelligent guiding systems for smartphones.
Year: 2010 - 2013
Senior Software Engineer
Overview:
While working at Ibertech, I performed roles as Senior Consultant and IT Trainer on areas such as software engineering and project management, working in big software companies such as INDRA, Everis, INSA, among others. As a consultant, I provided integrated solutions for the development of large projects in sofware industry, and as a IT Trainer I provided specialized training in areas such as Programming, Analisys & Design, UML modeling, Software Architectures, Test and Quality, SCRUM, etc.
Year: 2010 - 2012
Research Visitor
Details:
- Host: Imperial Collge London (UK)
- Laboratory: ISN - Intelligent Systems and Networks
- Field: Robotics and Machine Learning
- Project: ACBO
Overview:
During my visit to the Imperial College London (ISN -Intelligent Systems and Networks- Lab), I conducted research activities such as: design and implementation of a bio-inspired computational architecture for decision making with application on both simulated and real robots, running experiments, and writing scientific papers.
Year: 2009 - 2010
Test Manager
Overview:
I led a team of 15 people (testers, programmers, etc.) on tasks such as: design and execution of functional tests for security systems, development of plug-ins for testing, development and documentation of test suites and test cases (under TDD model), implementation of unit tests, and generation of automated/robotic tests using IBM Rational.
Year: 2006 - 2007
Software Architect & Team Leader
Overview:
I was responsible for: designing multi-layer architectures for web applications; defining guidelines for OO A&D based on best practices; estimating project's size, effort, complexity, budget, and schedule; project managent, task assignment, and planning; and leading & coaching a 30-people development team (usign SCRUM methodology).
Year: 2005 - 2006
Software Architect & Team Leader
Details:
- Employer: Hewlette Packard (Colombia)
- Sector: IT
- Project: EDS-CGR
Overview:
I was responsible for: defining a J2EE Architecture for a web applications, developing most of the critical architectural use cases, developing mechanisms for integrating legacy systems with web applications, developing Services and Workflows, collaborating with SPI (Software Process Improvement) activities, defining metrics and estimation methods, and collaborating with the certification process of CMMI 4.
Year: 2002 - 2005
Associate Teacher and Researcher
Details:
- Institution: Konrad Lorenz University (Colombia)
- Department: Computer Science
- Projects: SAB and MARA
Overview:
I was responsible for conducting research activities in topics such as Artificial Immune Systems, Machine Learning, Neural Networks, and Classifier Systems. Also, I was in charge of teaching lectures such as: Advanced Programming in Java, Software Engineering, and Artificial Intelligence.
Year: 2002 - 2007
Software Developer
Overview:
I was responsible for: developing a web enable application for the Price WaterHouse Coopers using a J2EE architecture, and performing integration tests.
Year: 2000 - 2002
Projects. (Click below)
InMind
Overview:
The InMind project is the next next generation of intelligent personal assistants, an initiative to develop a possible smarter version of systems such as Siri, Google Now, Cortana, etc.
My Contribution:
- Designing and Implementing a Cognitive Architecture for Intelligent Agents
- Development of Middleware for Pervasive Computing and Service Composition based on Machine Learning and sentence embeddings
Year: 2014 - present
Funded by: OATH - Verizon - Yahoo!
My Work, InMind News
SARA
Overview:
SARA (Socially-Aware Robot Assistant) is capable of detecting social behaviors in conversation, reasoning about how to respond to the user's intentions, and generating appropriate social responses while doing her task duties at the same time. SARA uses different kind of AI techniques to carry out things such as: task and social reasoning, NLP, Rapport estimation, verbal and non-verbal behavior detection, among others.
My Contribution:
- Implementation of a Dialogue Manager Module (Task and Social Hybrid Reasoners)
- Implementation of Emotion Recognition Mod.
- Impl. of Search and Recommendation System
Funded by: World Economic Forum 2016
Website
RCTA
Overview:
The RCTA project (Robotics Collaborative Technology Alliance) brings together government, industrial, and academic sectors in situations to address R&D required to enable the deployment of future military unmanned ground vehicle systems ranging in size from man-portables to ground combat vehicles.
My Contribution:
- Implementing a cognitive mission planner to decompose user-supplied commands into a mission plan composed of tactical behaviors
- Running Experiments and Writing Papers
Year: 2013 - 2014
Funded by: US Army Research Lab
Biennial Program Plan (pdf)
ACBO
Overview:
The main purpose of this project was to create an architecture to support cognitive processes such as perception, learning, decision making, planning, consciousness, and adaptive behaviors. Experiments were carried out on both real and simulated robots.
My Contribution:
- Design & Implementation of an Architecture to support high level cognitive processes
- Implementation of Hybrid Machine Learning and Evolutionary Decision-Making modules
- Impl. of a Behavior Orchestration Module
Year: 2009
Funded by: UPM and Imperial College London
Watch videos
SAB
Overview:
The purpose of this project is to simulate animal behaviors by using machine learning algorithms in an animat environment. Preys and Predators animats were simulated: they compete for resources, hunt each others, create coalitions, and exhibit behaviors led by instincts and drives.
My Contribution:
- Implementation of a hybrid system based on machine learning: Classifier Systems, NN, Genetic Algorithms, and Immune theory
- Running experiments and writing papers
Year: 2005 - 2007
Funded by: Education Ministry of Spain
Wath videos
MARA
Overview:
The main purpose of this project was to develop a bio-inspired subsumption architecture for robots. Every layer is controlled by a different machine learning algorithm that allows the robot to carry out tasks from obstacle avoidance to complex planning.
My Contribution:
- Implementation of hybrid machine learning algorithms: NN, Classifier Systems, Artificial Immune Systems, and Genetic Algorithms.
- Implementation of a simulated robotic environment
Year: 2003 - 2005
Funded by: Konrad Lorenz University
Watch videos
MILES-SEPIA
Overview:
Interaction Models Centered on Language, Space and Computational Semantics. This project combines dialogue systems, NLU, 3D virtual environments, and semantic representations. Ontology models allow reasoning about spatial concepts and their interaction with path finding and user's perception.
My Contribution:
- Implementation of an Intelligent Multi-Agent System that uses Machine Learning
- Implementation of Monitoring, Semantic, Trajectory and Communication Modules
Year: 2011 - 2013
Funded by: Education Ministry of Spain
Demo MILES-SEPIA
CHIL2
Overview:
This poject mainly focused on developing a collaborative platform (community users) that allows raw materials suppliers from European countries to connect with other suppliers, providers and companies by linking contents in an automatic and intelligent manner (using semantic web and intelligent tagging).
My Contribution:
- Designing a software architecture that supports Intelligent Tagging
- Designing and Implementing Ontology Models
- Implementing a large-scale semantic database
- Creating services for Inteligent Queries
- Scaling the system up
Year: 2013
Funded by: European Agriculture Commission
REHABILITA
Overview:
The REHABILITA project provides medical rehabilitation and therapy personalization over mobile devices. Research areas: intelligent monitoring biomedical devices, virtual environments, machine learning, hybrid bionic systems.
My Contribution:
- Develoment of an adaptive recommendation system based on psychological motivation rules and machine learning
- Implementation of a decision-making systems based on user preferences
Year: 2012 - 2013
Funded by: GMV Innovating Solutions
News App
Cognitive Modeling
Overview:
This project describes the possible structures that underlie the decision-making process of well-known Network Science experiments through the modeling of symbolic cognitive prototypes supported by a robust and complex cognitive architecture so-called ACT-R.
My Contribution:
- Implementing cognitive agents and decision-making models in ACT-R Cognitive Architecture
- Designing a Multi-agent architecture for network science experiments
- Running simulations and experiments
- Writing Scientific Papers
Year: 2013 - 2014
Funded by: US Army Research Lab
Alhambra & Partenon
Overview:
Alhambra is the Multichannel Architecture built on Corebanking Software, which offers an end-to-end process view for enhanced information use and integration. Partenón architecture provides a Banking system for the retail segment of the Santander Group in Europe and United States.
My Contribution:
- Implementation of Software Modules (Mira, Vega, WSAD) using J2EE Enterprise Architecture
- Team Manager: leading the Test Team, executing robotic testing
Year: 2006 - 2007
Funded by: ISBAN Santander Bank
WebsiteEDS-CGR
Overview:
This was the largest project on Latin America developed by Hewlette Packard. This is a corporate portal that allows access to all the departments of the Government Accountability Office (government agency that provides auditing, evaluation, and investigative services for the Republic of Colombia).
My Contribution:
- Design, validation and implementation of the Sofware Architecture
- Defining SPI practices for CMMI 4
- Leading a 30-people Development Team
Year: 2003 - 2005
Funded by: Government Accountability Office
WebsiteHispatec ERPagro
Overview:
ERPagro is the leading solution for companies in the agrifood industry. For over 25 years, they have developed specific technology for companies that export fruit, vegetables, horticultural products, cereals, oils and wines all over Spain and Europe.
My Contribution:
- Designing and Implementing the Web Application's Software Architecture
- Lead the A&D team (22 people)
- Support to the SPI process
Year: 2005 - 2006
Client: Hispatec
Working for: Unisys
WebsiteSubocol
Overview:
Subocol was a Web Portal developed jointly with Oracle Consulting and PricewaterhouseCoopers for the purchase of spare auto parts and vehicles via Internet.
My Contribution:
- Implementation of Software Modules under the J2EE Enterprise Architecture)
- Conducting Software Tests and Validation
- Reverse Engineering of models
- OO Analysis and Design
Year: 2002 - 2003
Funded by: PricewaterhouseCoopers
Working for: Oracle
Website
[R&D]
2014 - Present
[R&D]
2016
[R&D]
2013 - 2014
[R&D]
2009 - 2010
[R&D]
2005 - 2007
[R&D]
2003 - 2005
[R&D]
2011 - 2013
[R&D]
2013
[R&D]
2012 - 2013
[R&D]
2013 - 2014
[Industry]
2006 - 2007
[Industry]
2003 - 2005
[Industry]
2005 - 2006
[Industry]
2002 - 2003
Selected Publications.
- Hybrid Architecture for artificial Cognitive Systems which enact emergent, adaptive and self-organizing behavior. PhD. Thesis, 2011 (Spanish). [Document] [Presentation] [Book]
- Hybrid Machine Learning Architecture for Virtual Agents. MSc. Thesis, 2007 (Spanish).
- Intelligent Tutoring System for Mathematics that uses an Evolutionary Student Model. BSc. Dissertation, 2002 (Spanish). [Document] [Presentation] [Book]
- Synergistic Integration of Large Language Models and Cognitive Architectures for Robust AI: An Exploratory Analysis. AAAI 2023 Fall Symposium on Cognitive Architectures and Generative Models [Download]
- Propositional Reasoning via Neural Transformer Language Models. NeSy 2022[Download]
- A Task-Oriented Dialogue Architecture via Transformer Neural Language Models and Symbolic Injection. SIGDIAL 2021[Download]
- A Long-Term Evaluation of Adaptive Interface Design for Mobile Transit Information. MobileHCI 2020[Download]
- Dynamic Service Composition Orchestrated by Cognitive Agents in Mobile & Pervasive Computing. IEEE Services 2019 [Download]
- NLSC: Unrestricted Natural Language-basedService Composition through Sentence Embeddings. IEEE SCC 2019 [Download]
- Cognitively-inspired Agent-based Service Composition for Mobile & Pervasive Computing. AIMS 2019 [Download]
- InstructableCrowd: Creating IF-THEN Rules for Smartphones via Conversations with the Crowd. Human Computation 2019 [Download]
- CogArch-ADL: Toward a Formal Description of a Reference Architecture for the Common Model of Cognition. Procedia Computer Science Journal 2018 [Download]
- Higher-level Knowledge, Rational and Social Levels Constraints of the Common Model of the Mind. Procedia Computer Science Journal 2018 [Download]
- Metacognition for a Common Model of Cognition. Procedia Computer Science Journal 2018 [Download]
- SOGO: A Social Intelligent Negotiation Dialogue System. IVA 2018 (In Press) [Download]
- An Efficient Mobile-based Middleware Architecture for Building Robust, High-performance Apps. ICSA 2018 [Download]
- Cognitive-Inspired Conversational-Strategy Reasoner for Socially-Aware Agents. IJCAI 2017 [Download]
- An Architectural Approach for Developing Intelligent Personal Assistants Supported by NELL. Encontro Nacional de Inteligência Artificial e Computação [Download]
- Socially-Aware Animated Intelligent Personal Assistant Agent. SIGDIAL, 2016. [Download]
- Integrated Intelligence for Human-Robot Teams. International Symposium on Experimental Robotics, 2016. [Download]
- Strong CogSci: Guidance from Cognitive Science On the Design of a Test of Artificial Intelligence. AAAI 2015 Workshop: Beyond the Turing Test.[Download]
- Cognitive Modeling of Behavioral Experiments in Network Science. AAMAS, 2014. [Download]
- Simulating Network Behavioral Dynamics by using a cognitively constrained Multi-agent approach. BRIMS, 2014. [Download]
- An evolutionary behavioral model for decision making. Adaptive Behavior Journal, 2012. [Download]
- Evolving the way of doing the right thing. IEEE CEC, 2012. [Download]
- Modulation of multi-level evolutionary strategies for artificial cognition. GECCO, 2009. [Download]
- Hibridization of cognitive models using evolutionary strategies. IEEE CEC, 2009. [Download]
- Hybrid Behaviour Orchestration in a Multilayared Cognitive Architecture using an Evolutionary Approach. IEEE WCCI, 2008. [Download]
- Bio-inspired Cognitive Architecture for Adaptive Agents based on an Evolutionary Approach. AAMAS, 2008. [Download]
- Analysis of Emergent Properties in a Hybrid Bio-inspired Architecture for Cognitive Agents. HAIS, 2007. [Download]
- Multilayered Evolutionary Architecture for Behaviour Arbitration in Cognitive Agents. EL Journal, 2007. [Download]
- Intelligent Tutoring System using an Evolutionary Student Model. IASTED, 2002. [Download]
Contact.
