New-Task Planning in Robotics

Abstract

The motivation behind this project was to investigate methods of how it could be possible to increase the autonomy of a robot through new-task planning and re-planning. This thesis addresses this investigation by firstly, researching the existing literature. This research gave a foundation of what research is currently being done and what have been successful in the past. After the foundation was set, the project moved over to a development phase where a simulated environment was setup with different coloured blocks of different various shapes. In this environment, a simulated TIAGo robot was inserted. After the simulated environment was setup, the work progressed into a phase of developing a program that used case-based reasoning to make the robot be able to retrieve similar past cases based on attributes when it was faced with a new task. In testing, it was revealed that the method was successful, and the robot was able to execute single-action task execution and combination task execution. In combination task execution the robot was able to string together several cases into a completely new task in order to solve more complex tasks than that of single-action tasks execution.