The goal of this project is to understand how groups of robots accomplish tasks by working as a team. Our first task is to design algorithms that will enable a team of robots to visually inspect and understand an object, such as a car or a tree. Equipped with cameras and other sensors the robots will, just like a team of students, exchange information with each other and in this process form a coherent model of the observed object. With this model the robots can then move on to explore the environment and model new or recognize already seen objects.
In the process of modeling the information exchange we will study how the data flows between individual robots and how it should be optimally exchanged to lead to a consensus. We will also examine situations where the consensus cannot be formed and try to understand why this happens. The core of the computational modeling will be based on our recent work on distributed learning and probabilistic inference. We will also study how this collaborative learning among robots can be used to model learning and information exchange in human social network.
We will use teams of Rover robots (small tank-like robots) equipped with cameras, microphones, and IR illuminators. Adventurous students will also get a chance to play with our quad copters.
Position type: Five research hours per week or more
Period: May 2012 on
Application deadline: ongoing
- Major: Computer Science, Electrical Engineering
- GPA: GPA of 3.5 or higher
- Knowledge of C/C++/Matlab or Python
- Advanatge: experience with OpenCV / ROS / OpenNI
To apply contact Prof. Pavlovic at vladimir _at_ cs.rutgers.edu.