The thesis “ROILA – Robot Interaction Language” has been printed. Please join us for the defense on June 1st, 16:00h in lecture theater 4 of the Auditorium at the Eindhoven University of Technology. This will be followed by a reception in Senaatszaal.
You are currently browsing the archive for the News category.
Good news! We received a small research grant from NZILBB to create full course book about ROILA. This book is intended to help students to learn ROILA. Dean Sutherland will help with supervising a student that can create this course book.
At this moment, we are documenting the entire ROILA project. Soon we will provide this information. It will also include details about the ROILA curriculum and the controlled experiment that we conducted at the Huygens College Eindhoven.
Tags: ROILA documentation
Recently we concluded the first ever ROILA course at the Huygens College Eindhoven. The course lasted for three weeks and was taught to 100 Dutch high school students students spread over 4 classes. The lessons involved practicing ROILA and playing with LEGO robots. All of us (teachers and students included) had alot of fun. So much so the school wants to make it a permanent part of their Science curriculum. They plan to do a repeat of the course for another class in January next year. A follow-on controlled experiment was also conducted which compared the efficacy of learning ROILA to control a robot vs controlling English-savvy robot. ROILA was found to better on both subjective and objective accounts (details coming soon).
We would like to thank the staff and teachers of Huygens College Eindhoven for their support and appreciation. Needless to say the children deserve a special mention too.
We will be continously updating the ROILA academy section as we progress with the ROILA course at Huygens College, Eindhoven. Please visit the academy to know more about the course and our efforts at the school.
The first set of 10 ROILA lessons are now uploaded in the ROILA academy. This set is the first version and may be subject to slight revisions. Please feel free to join the ROILA academy which is setup in Moodle. Also uploaded is the relevant glossary for these lessons, both here on the website as well as in the academy on Moodle. Alternatively you might want to have a look at the HTML version of the lessons. We would like to acknowledge the contributions of Jerry Muelver towards the design of the lessons. Please let us know if you have any comments or would like to give us some feedback.
The entire ROILA audience is welcome to study the ROILA lessons. If you are interested in subscribing to the academy please send us an email with your name and location so that we can make an account for you.
We are busy implementing small scenarios/concepts for students of ROILA so that they can interact with Mindstorm NXT robots while talking in ROILA. One of them is a simple shooting game, where we wanted to get the color sensor to work. However the “new” color sensor in Mindstorms NXT 2.0 does not work with lejos NXJ 0.8 and neither with some versions of 0.85. Please use the following classes.jar file to make your sensor work, courtesy of the lejos team. Remember to replace your old version of the jar file in two places: lib folder in the projects folder and lib folder where you installed lejos. A simple replace worked for us and we did not have to rebuild any code. But, remember to download your java program on the NXT again.
At this moment we are busy designing lessons for ROILA. These lessons will be used by high school students to learn ROILA in our upcoming project. We are using Moodle an open source e-learning tool. Moodle provides various functionalities and features which make designing lessons fun and simple. One of the interesting challenges that we are contemplating is how to provide a dictionary or a glossary with audio pronunciations of ROILA words. An alternative could be to accomplish this externally using word press on the ROILA website (see first version). We aim to use Text to Speech recordings as the sample pronunciations for ROILA as that would be consistent and efficient. The recognition accuracy of the TTS sentence recordings was found to be very high (just 1 word error in about 120+ words) for a list of 30 sentences. This figure was generated by passing recordings offline in the Sphinx-4 recognizer.
We are extremely pleased and grateful that Christiaan Huygens College Rachmaninowlaan in Eindhoven has agreed to cooperate with us in the ROILA project. Children from third grade (14-15 years old) will be learning ROILA and using it to play games with LEGO robots. This activity will be part of their Science Class and we expect the ROILA learning to last for 3-4 weeks with a final one week used for the evaluation and game play. As a token of appreciation we will be donating LEGO Mindstorms NXT kits to the school.
We are extremely pleased to welcome two additional members to our team who will be helping us in this project. Hanneke Hooft van Huysduynen (Final Year Bachelor Student at Industrial Design, Eindhoven University of Technology) will be helping with the design of the game(s) and the execution of the study. Jerry Muelver (President of North American Ido Society, Inc) is assisting us by designing the ROILA lessons and lectures. We welcome his expertise with artificial languages and conlangs.
We will post updates and details as soon as they become available.
Ubuntu users will find it quite easy to install and use Festival for Text to Speech. The following link gives the details:
We would strongly recommend to use Ubuntu to run TTS. Compiling and building Festival on Windows is quite a task. All of this means that ultimately we are faced with two choices when implementing a complete application for ROILA, since we had configured the Lejos NXT platform on Windows.
- Setup lejos on Ubuntu, including Bluetooth support.
- Use the text2wave utility in Festival to generate wav files on Ubuntu. These wav files can then be played from within Java when running NXT programs on Windows.
We have prepared a simple script that generates wav files of any number of sentences written in a text file using approach 2. Of course the disadvantage is that TTS is done offline and one would need to know beforehand the plausible list of sentences that the NXT robot could say. If someone has hands on experience with approach 1 please let us know.