2012 Technogenesis Projects - Systems and Enterprises
Front-End Web Developer for Threat Detection Applications
Looking for student interested in developing the front-end of an application to design inspection strategies for threat detection at container terminals and airport security checkpoints. The application is based on an evolutionary algorithm that is currently coded in VBA.
Understanding of mathematical optimization methods is preferred. Emphasis will be made on the graphical representation of results.
Phase I: Research background understanding On Week 1 and 2 the student will review a suggested list of research papers to understand the background and importance of the software tools/he will create. The list includes but is not limited to three papers on container inspections and one paper on passenger screening.
Phase II: Evolutionary algorithm understanding On Week 3 the student and mentor will discuss the evolutionary algorithms presented in the suggested list of research papers, all already coded in VBA. By the end of Week 3, the student is expected to understand the back-end of the software tool and propose a work plan for the front-end development phase. The plan must involve integrating/adapting the current back-end code to the front-end.
Phase III: Developing, Verification and Validation From Week 4 to Week 9, the student will follow his/her proposed work plan to develop a user-friendly software tool. Along the process, the student can verify his/her code using results reported by the VBA code.
Phase IV: Documentation On Week 10 the student will write a user manual to describe how the tool works.
Advisor: Prof. Jose E. Ramirez-Marquez, PhD
School of Systems & Enterprises
Systems Engineering Research Review
Task: Preparation and synthesis of SERC research, consolidation of research into annual research review, include collaborative communication.
Preferred qualifications: good writing and communication skills; familiarity with Word/Excel/Access; any discipline OK; interest in systems or systems thinking.
Advisor: Deborah Factor Lepore
School of Systems and Enterprise
Positions available: 2
Voice Command and Natural Language Modeling to Support Test Generation for Verification of Self-Adapting System
The future vision is to allow people to directly communicate through voice commands using natural language to support real-time testing of self-adapting systems. Specifically, we'd like to be able to allow humans (at least in the near term) to be able to communicate with self-adapting robots and help guide or check after the fact that the self-adapting behaviors are correct given environmental and operational constraints. This would be applicable to many applications where humans interact with robotic or unmanned systems:
- Soldiers in the field directing unmanned systems - People directing robot operations on drilling platforms
- Robots conducting maintenance on the future Smart Grid
- Robotic surgery
This topic is inspired by three ideas:
- Siri is a voice command assistant for the iPhone 4, but Apple has chosen to not allow it to be used to create programs, at least for now.
- In collaboration with a group of researchers in Brazil, who have converted a constrained natural language into a modeling language originally created by the Naval Research Labs called SCR that is now supported by a tool called the T-VEC Tabular Modeler (TTM). Therefore, we know that we can convert a constraint natural language into SCR.
- SCR is a declarative language, which makes it easier to create than say a programming language such as Java.
- TTM integrates with a test vector generation tool.
In collaboration with our Stevens robotics researchers, we're looking at embedding models and verification (e.g., testing) mechanism within autonomously self-adapting robots.
The projects available are:
Section 1 (2 student positions available)
- Investigate using open source software to process voice commands - the voice commands will logically move through the model to place process Natural Language (NL) Text.
- Transforming natural language, for example using Apache OpenNLP (a machine learning based natural language processing toolkit)
- Creating simple models which can be imported to TTM to generate tests
Section 2 (2 student positions available) - possibly in conjunction with Drs. Cappelleri and Zavlanos
- Investigate how small robot adaptations can be tested in a running robot
- Using a robot simulator to test hypotheses
Advisor: Prof. Mark Blackbuurn
School of Systems and Enterprises