Robocup Project Proposal

Ethan Leland
Mert Akdere

1. Abstract

In our project we will investigate the Robocup Soccer domain. We will be using Sony AIBO Robots as our hardware and OPEN-R SDK for software development in this platform. We will build a basic application in which two AIBO robots will pass the ball to each other and finally score a goal on a soccer field while avoiding obstacles on the playing field. The overall intention is to apply these techniques and gathered experience to compete in the Robocup Competition.

2. Detailed description of the application

Robocup is a multi-robot environment. Multi-Robot teams with teammates that can communicate can build more accurate world models, cooperate better and act more rationally. Yet in practice building such a consistent world model and a functional robot team is difficult. Some of the problems are due to communication latencies, merging information from different sources and noisy data.

In this project through our sample application, we will analyze:

• how data is gathered on AIBO robots
• what are the different sensors that provide data
• how to construct consistent world models using collected data
• what kind of data is relevant for communication between robots
• how to disseminate data while taking constraints such as latency and priority into account

We will be applying localization techniques using data collected via low level vision processing to build a consistent world model over two Aibo robots, then use this information to create a policy for the robots to implement. We will be using this world model to implement a team play strategy. Some current ideas include using potential fields and stream processing to have successful collaboration between robots.
Target users for this project include the new Robocup team at Brown as well as other robocup teams who are interested in advanced team play.

3. Motivation

Developing an application for competing in the Robocup competition involves combining work from different areas of computer science. Therefore, the Robocup Competition is the center of attention in computer science research.
Here, in Brown, a robocup team has recently been formed. The team has a goal to participate in the Robocup Competition this spring. Hopefully our efforts will help the team achieve their goal.

4. Methodology

As stated, we will be using the Aibo robots and the Open-R programming environment For motion, we are borrowing techniques from other well-established robocup teams such as UPenn and CMU. Our localization will be based on a Monte Carlo Localization routine implementing a particle filter. To achieve the low level vision, we will have very distinct markers in the field that we will be able to identify through simple vision techniques, which will aid in our localization.
The Open-R environment that we will be programming in is a C++ based framework provided by Sony with very specific guidelines that we will first have to learn in order to complete this project.

5. Timetable and workload division

Teleoperation	10/27
Communication by Mert	Motion by Ethan
Basic Vision and Localization	11/20
Vision by Mert	Localization by Ethan
Implementation Logic	12/1
Division of work is TBD

6. Related work

Some Robocup participants include Upenn, CMU, and the German Team. We are currently studying the techniques that these teams used and will be incorporating them into our own project. We think that our project will go further into some of the aspects of team play that we feel are missing in some of these current Robocup teams.