Difference between revisions of "Intrusion Detection in Swarm Robotics"

From REU@MU
Jump to: navigation, search
 
Line 2: Line 2:
  
 
'''Mentor:'''  [http://www.marquette.edu/mscs/facstaff-perouli.shtml Dr. Debbie Perouli]
 
'''Mentor:'''  [http://www.marquette.edu/mscs/facstaff-perouli.shtml Dr. Debbie Perouli]
 
 
'''Description:''' When robots are organized in swarms, each robot coordinates with the rest in order to execute a collaborative task. Potential applications vary significantly and include warehouse cleaners, space explorers and nanorobots in veins and arteries. Although fault tolerance algorithms have been developed to some degree for such swarms, there are important security questions that remain open. This project focuses on detecting whether one or more members of the robotic swarm have been compromised. Building on preliminary work, the student(s) will develop, implement and test an intrusion detection algorithm for specific robotic swarms. The student(s) will be able to work with both simulators and real robots available at Marquette University’s research laboratories.
 
  
 
==Project Description==
 
==Project Description==

Latest revision as of 17:49, 1 June 2018

Student Researcher: Lindsey Coffee-Johnson

Mentor: Dr. Debbie Perouli

Project Description

Several industries, such as healthcare and hospitality, are investigating the benefits of using social and service robots to cover some of their customers' needs. For example, a robot could interact with an older adult to decrease the feeling of loneliness or a robot could deliver items such as food or towels to a hotel guest. Successful deployment of social and service robots will likely lead to the introduction of more robotic units in the facility resulting in the formation of a swarm.

The objective of this project is to develop techniques that detect a hacked robot in the context of a small swarm (three to four robots). The project will focus on two types of abnormal robot behavior:

  • The robot's location is different from the location it should have. This type of abnormality includes the robot spending significantly more or less time at a location that it was supposed to visit.
  • The robot's psychological impact on the customer is negative.

Milestones and Goals

Week Description
Week 1: Orientation
  • Familiarize with Marquette, project and mentor
  • Set milestones and goals for the project duration
  • Start reading related literature on trust in the context of human-robot interaction and intrusion detection
Week 2: Study Related Work
  • Read related literature with an emphasis on cognitive science topics related to persuasion, deception and manipulation, as well as security in cyber physical systems (CPS) and fault tolerance.
Week 3: Learn About TurtleBots
  • Learn how to program TurtleBots
  • Explore the Robot Operating System (ROS) and simulator
  • Install and run Python programs developed by Dan Cronce during previous REU project
Week 4: Form Research Hypothesis
  • Specify assumptions
    • Tasks that each robot will perform (normal behavior)
    • Acts that will constitute proof of hacking (abnormal behavior)
Week 5: Design Methodology to Support or Disprove Hypothesis
  • Experiments for the part of the project related to hacking the robot's location
  • Comparisons to existing cognitive science studies for the part of the project related to the psychological impact on consumer
  • Give midway presentation
Week 6: Start Writing Paper
  • Complete the first part of the paper
Week 7: Build System
  • Build system needed to run the experiments with the TurtleBot 2 robots
  • Collect any additional related work evidence
Week 8: Run Experiments
  • Run experiments on TurtleBots
  • Detail comparisons of the suggested technique for identifying intrusion detection to existing literature
  • Prepare final poster on research
Week 9: Test hypothesis
  • If needed, rerun experiments
  • Consider future work
Week 10: Presenting Research
  • Present at poster session
  • Prepare and give oral presentation
  • Finish and submit final paper