ROS-BASED INDOOR SERVICE ROBOT WITH NAVIGATION, SPEECH RECOGNITION AND IMPROVED FACE RECOGNITION MODEL

• Main Author: Ali Shabbir
• Format: text::Thesis
• Language: English
• Published: 2023
• Subjects: ROS-BASED INDOOR SERVICE ROBOT WITH NAVIGATION, SPEECH RECOGNITION AND IMPROVED FACE RECOGNITION MODEL

The use of service robots is more advent in the current years. In many scenarios, these robots have supported humans and eased various day to day tasks. Service robots are often used to transfer objects from one point to another. However, service robots lack the ability to identify human, specifically if the user wishes to deliver to a target individual. Furthermore, most service robots are built on various platforms, thus, difficult to reinvent robots without changing the platform and system framework. Therefore, this work adopts a Robot Operating System (ROS) framework for service robot ROS, which is adaptable and easy to make relevant adjustments. This study designs a ROS based service robot navigation with face recognition for path generation that has the ability to recognize the user and deliver the package to the target individual. A major challenge in this application was choosing the precise face recognition model, which is compatible to ROS, as the built-in model of face recognition in ROS is neither accurate nor fast. Therefore, this study implemented an open source face recognition model and made it compatible with ROS. To explain the accuracy of both models, the face recognition model was compared with the built-in ROS visualization model. Finally, a TurtleBot mobile robot platform was used as a service robot that works on ROS platform with integrated face recognition model and navigation package. The series of experiments was conducted in a simulated office environment as well as implemented in simulation and real-world experiments were consistent and proved the successful implementation of face recognition in mobile robot path planning and navigation.