Social / Educational Robot to Improve Linguistic Skill for Toddlers Aged 5 to 6 Years old

Chew, Yi Seng (2018) Social / Educational Robot to Improve Linguistic Skill for Toddlers Aged 5 to 6 Years old. Final Year Project (Bachelor), Tunku Abdul Rahman University College.

Text CHEW YI SENG - Full Thesis.pdf Restricted to Registered users only Download (2MB)

Abstract

The purpose of this project is to design a social/educational robot to improve toddlers aged 5 to 6 years old linguistic skill. English is taught by the robot because the English standard in Malaysia is kept dropping especially for non-English speaking background family. In order to design a robot, aluminium profile has used to build the structure of robot based on the normal height of 5-6 years old toddler. These will easier the robot to interact with toddlers. On the other hands, the robot arms are printed out by using 3D printer and assembly with the structure of robot. RC servo motors are used to combine with the robot arms to provide motion. There will be body motion performs to interact with toddlers and attract attention of toddlers. For the facial expression of the robot, it consists of three different type of smiley face because smile on others can make people smile back. Raspberry Pi 3 is selected as the microcontroller because it is able to connect Wifi and support with multiple open source software. The speech recognition software platform used is Jasper. Therefore, the robot will consist of social capabilities and educational capabilities. Jasper is selected due to having multiple choices of text to speech (TTS) engine and speech to text (SST) engine. Furthermore, Jasper has simple interface compared to other speech recognition platform. For the TTS engine, SVOX Pico is selected because it works offline, sound clear and speaks with female voice. Female voice is used to communicate with toddlers because female voice is more soothing compared to male voice and it makes people feel comfortable. SVOX Pico TTS engine will convert programmed text to speech and tell the users. Pocketsphinx STT engine was chosen due to it works offline. It will receive the speech from users and convert it to text for the speech recognition software platform to understand. Four English words are selected to teach toddlers after refer to 3Q MRC Junior standard. These four words are happy, clap, stamp and shout. Other than teaching, robot will play song with these four words for toddlers. Two experiments have been conducted for both teacher and social/educational robot. The first experiment is focused on training and pre-test these four words for toddlers. Both teacher and robot will conduct the first experiment and compared based on the total average time took toddlers to pronounce. Second experiment is conducted to test whether toddlers are learnt these four words. This experiment is compared based on the average time took toddlers to pronounce and the success rate of pronunciation. Both experiments and comparisons are analysed in order to determine whether the social/educational robot able to improve toddlers 5 or 6 years old linguistic skill or not. For the first experiment of word clap, stamp and shout, the total average time taken toddlers to pronounce for social/educational robot is lower compared to total average time took toddlers to pronounce for teacher except the word happy. The average time taken toddlers to pronounce for second experiment is almost the same between social/educational robot and teacher. For teacher, the success rate toddlers able to pronounce is lower compared to social/educational robot.