Türk işaret dili için insansı robotlar üzerinde vücutlandırma çalışmaları

Visualizing real world in computer simulations has an essential role in analysing the complex situations. Working with physical devices can be hard and expensive because of unwilling damages and costs. Simulating cases and actions improve the success of projects. Especially in robotic projects, simulators have vital role to prevent accidental damages. Simulators also provide low cost training environments, so more people can study on it. Humanoid robots are widely used in education area and also for therapy of autism. Researcher try to find a common ground to communicate with hearing impaired people and people who have autism. Sign Language provide a common layer for people who can not speak verbally. This work is part of an ongoing project which aims to actuate humanoid robots as sign language tutoring assistance. Main goal of the project is to help children with communication problems, particularly children with Autistic Spectrum Disorder and hearing impaired children. Our Humanoid robot Robovie R3 does not have a 3D simulator, so this study mainly focus on creating a 3D simulator for our robot and examining success of it on performed sign language words. There are lots of developed physic engines and simulation environments to visualize projects in computers. Gazebo simulator is an open source improved one and provides many futures. Modelling actions in simulators and transferring them safely to physical world increase success and decrease actual time in projects. Benefits of simulators and necessity of creating Robovie R3 simulator motivate us to develop a simulator on Gazebo environment. It is capable of moving arms and head which is basically enough to sign TSL (Turkish Sign Language) words. In this study, two sub projects are developed. First one is Robovie R3 simulator on Gazebo environment. It can create motion files and convert them to transfer physical robot. TSL words are performed and saved as motion files on Robovie R3 simulator. In this way both a virtually embodied 3D R3 robot can be created, and also its actions can be transferred to a physical embodied R3 robot. Other project is an android based application, which provide a survey and test environment to examine the quality of showed signs on Robots and simulator and by that means, can serve as a tool to compare and test the effects of different embodiments of the humanoid robots in sign language tutoring. It can be modified to collect data in other projects. The overall scenario includes three experiment modes. In first mode, TSL words are created in Roboive R3 simulator and played in 3D environment. Collisions on signs are seen and they are fixed to protect physical robot. Performed TSL words are saved as motion files. In second mode, saved files converted to Roboviemaker2 format which is proper to play on physical robot. Than these signs transferred to physical Robovie R3 robot and analysed to measure the success of Robovie R3 simulator. In last mode, performed signs are recorded on simulator environment and added to android based game environment. This game is played by hearing impaired primary school students which have advanced sign language skills. They solved the sign test and results are quite promising and motivate us for future studies.