Tracking system for a soccer robot game using neural network

Date of Publication

2001

Document Type

Master's Thesis

Degree Name

Master of Science in Computer Science

Subject Categories

Computer Sciences

College

College of Computer Studies

Department/Unit

Computer Science

Thesis Adviser

Florante Salvador

Defense Panel Chair

Elmer P. Dadios

Defense Panel Member

Remedios Bulos
Charibeth Ko

Abstract/Summary

The conventional tracking algorithm lacks the capability to learn. Approaches like the use of neural network, which has learning capability, may be incorporated to the tracking algorithm to take advantage of previously estimated pose. Neural network approach may be investigated in terms of speed and accuracy by comparing it with the conventional tracking algorithm. This research develops a pose estimation algorithm using neural network as its paradigm. Pose estimation is concerned with finding an object's position and orientation. There are several approaches in handling pose estimation, and one of them is through the use of neural network. Neural network, with its learning capability, can take advantage of previously estimated pose and use this for future estimation. The pose estimation algorithm will be tested in the game of soccer robots, specifically the Micro-Robot World Cup Soccer Tournament (MiroSot).

Abstract Format

html

Language

English

Format

Print

Accession Number

TG03228

Shelf Location

Archives, The Learning Commons, 12F Henry Sy Sr. Hall

Physical Description

1 v (various foliations) ; 28 cm.

Keywords

Neural networks (Computer science); Robotics; Games; Soccer; Computer vision; Robot vision; Micro League Football (Game)

This document is currently not available here.

Share

COinS