IMPLEMENTATION INDONESIA LANGUAGE SPEECH RECOGNITION AND FEED-FORWARD CONTROL FOR A SOCCER ROBOT
Since 2015, Bandung Institute of Technology has developed wheeled soccer robot. From 2016 to 2018, the robots have succeeded in joining Indonesian Robotic Competition on Wheeled Soccer Robotics Division (KRSBI Beroda). Bad performance of robot caused limited game play in competition. Biggest problem...
Saved in:
Main Author: | |
---|---|
Format: | Final Project |
Language: | Indonesia |
Online Access: | https://digilib.itb.ac.id/gdl/view/43432 |
Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
Institution: | Institut Teknologi Bandung |
Language: | Indonesia |
id |
id-itb.:43432 |
---|---|
spelling |
id-itb.:434322019-09-27T09:02:31ZIMPLEMENTATION INDONESIA LANGUAGE SPEECH RECOGNITION AND FEED-FORWARD CONTROL FOR A SOCCER ROBOT Sumpono Putra, Calmantara Indonesia Final Project Feed-forward, Kaldi toolkit, PID, Speech Recognition, Soccer Robot INSTITUT TEKNOLOGI BANDUNG https://digilib.itb.ac.id/gdl/view/43432 Since 2015, Bandung Institute of Technology has developed wheeled soccer robot. From 2016 to 2018, the robots have succeeded in joining Indonesian Robotic Competition on Wheeled Soccer Robotics Division (KRSBI Beroda). Bad performance of robot caused limited game play in competition. Biggest problems which often occurs are oscillating movement and localization also overshoot movement and localization. Best strategy can be achieved by reducing or eliminating overshoot and oscillation in the control responses. Refer to Robocup’s rules on World Soccer Robotics Competition, the additional features of speech recognition will be developed to improve the robot’s performance. The movement problems can be overcome by selection process of locomotion controller which is very important on the robot performance. Feed-forward controller is often used to increase the response systems towards set point changes. This controller is chosen to improve the robot movement. For speech recognition systems, Kaldi toolkit was chosen, which is commonly used by researcher in speech recognition technology and in professional. To sum up, feed-forward controller has a significant impact in terms of performance response times that are 200 milliseconds faster, settling time are 3 second faster, and the systems not oscillating at the end of target points. The implementation of speech recognition systems shown a good performance with Word Error Rate (WER) is about 7.6%. Testing for 105 inputs, provides systems accuracy is about 29%. The robot behavior shown correspond with speech command. text |
institution |
Institut Teknologi Bandung |
building |
Institut Teknologi Bandung Library |
continent |
Asia |
country |
Indonesia Indonesia |
content_provider |
Institut Teknologi Bandung |
collection |
Digital ITB |
language |
Indonesia |
description |
Since 2015, Bandung Institute of Technology has developed wheeled soccer robot. From 2016 to 2018, the robots have succeeded in joining Indonesian Robotic Competition on Wheeled Soccer Robotics Division (KRSBI Beroda). Bad performance of robot caused limited game play in competition. Biggest problems which often occurs are oscillating movement and localization also overshoot movement and localization. Best strategy can be achieved by reducing or eliminating overshoot and oscillation in the control responses. Refer to Robocup’s rules on World Soccer Robotics Competition, the additional features of speech recognition will be developed to improve the robot’s performance.
The movement problems can be overcome by selection process of locomotion controller which is very important on the robot performance. Feed-forward controller is often used to increase the response systems towards set point changes. This controller is chosen to improve the robot movement. For speech recognition systems, Kaldi toolkit was chosen, which is commonly used by researcher in speech recognition technology and in professional.
To sum up, feed-forward controller has a significant impact in terms of performance response times that are 200 milliseconds faster, settling time are 3 second faster, and the systems not oscillating at the end of target points. The implementation of speech recognition systems shown a good performance with Word Error Rate (WER) is about 7.6%. Testing for 105 inputs, provides systems accuracy is about 29%. The robot behavior shown correspond with speech command. |
format |
Final Project |
author |
Sumpono Putra, Calmantara |
spellingShingle |
Sumpono Putra, Calmantara IMPLEMENTATION INDONESIA LANGUAGE SPEECH RECOGNITION AND FEED-FORWARD CONTROL FOR A SOCCER ROBOT |
author_facet |
Sumpono Putra, Calmantara |
author_sort |
Sumpono Putra, Calmantara |
title |
IMPLEMENTATION INDONESIA LANGUAGE SPEECH RECOGNITION AND FEED-FORWARD CONTROL FOR A SOCCER ROBOT |
title_short |
IMPLEMENTATION INDONESIA LANGUAGE SPEECH RECOGNITION AND FEED-FORWARD CONTROL FOR A SOCCER ROBOT |
title_full |
IMPLEMENTATION INDONESIA LANGUAGE SPEECH RECOGNITION AND FEED-FORWARD CONTROL FOR A SOCCER ROBOT |
title_fullStr |
IMPLEMENTATION INDONESIA LANGUAGE SPEECH RECOGNITION AND FEED-FORWARD CONTROL FOR A SOCCER ROBOT |
title_full_unstemmed |
IMPLEMENTATION INDONESIA LANGUAGE SPEECH RECOGNITION AND FEED-FORWARD CONTROL FOR A SOCCER ROBOT |
title_sort |
implementation indonesia language speech recognition and feed-forward control for a soccer robot |
url |
https://digilib.itb.ac.id/gdl/view/43432 |
_version_ |
1821998873894191104 |