Skip to main content

Tools for Assured Autonomy

Primary Investigators:
Gabor Karsai
Abhishek Dubey 

Brief Description of Project:
Autonomous vehicles (cars, drones, underwater vehicles, etc.) started using software components that are built using AI and machine learning techniques. These vehicles must operate in highly uncertain environments and it is very difficult to design a correct algorithm for all situations. Instead, engineers collect data from a real or simulated environment and train a general purpose system - typically a neural net - to perform a certain function using machine learning techniques. The challenge is that the training data cannot cover all possible cases, yet one needs assurances that the system works safely and has acceptable performance.

This project is about fundamental research and the construction of design tools for the engineering of such systems. The tools help in modeling the system (e.g. an underwater vehicle), executing the training and testing the learning-based software components, and building formal arguments (called assurance cases) to show that the system is safe.

Desired Qualifications:
CS/CmpE/EE background with familiarity with concepts and techniques of signals and systems, computer architecture, software design, and embedded systems. Knowledge of the Python/C/C++ languages is a plus, as well as experience with ROS (the Robot Operating System).
Nature of Supervision:
The intern will work in a team of researchers and staff engineers on the project, under the supervision of a researcher. The intern will be given weekly assignments and report on those in project meetings.
A Brief Research Plan (period is for 10 weeks):
Week 1: getting familiar with the project
Week 2-4: Learning about the specific tools used and being developed on the project
Week 5-9: Contributing to the tool development addressing a specific research problem
Week 10: Preparing a final report on the research results

Number of Open Slots: 1
Contact Information:
Name: Gabor Karsai
Department: Computer Science