Towards interactive sketching robots
: sketch classification and generation

  • M. U. Bigumjith Dias

    Student thesis: Master's Thesis

    Abstract

    This thesis proposes a novel sketching robotic system that interacts with the user in realtime. Sketching is not a mere aesthetic skill but helps to develop the cognitive abilities of human beings. Sketching is practiced in schools and other educational institutes using teacherdirected traditional teaching methods. This thesis proposes a human–robot sketching system using the state-of-the-art robot and machine learning models. The human–robot interaction can motivate users to learn sketching techniques. Individual interaction with each user will yield productive teaching. This robotic platform can be introduced to schools and institutes due to its accuracy, cost-effectiveness, and affordability. The new system is employable and affordable to existing educational institutions. Therefore, the proposed sketching robot is simple, efficient to train, and can improve its performance over time. On the other hand, the robot can interact with the user and produce new sketches to demonstrate specific sketching techniques. This research answers several research questions, such as “how to recognize the topic of interest of the user?” and “how to generate novel sketches associated with the topic of interest?”. Users have diverse perspectives and interests; therefore, the sketching robot can recognize the user’s topic of interest and generate related sketches accordingly. In particular, the system applies novel machine learning models to recognize the user’s topic of interest, i.e., sketch classification, and render comparable sketches, i.e., sketch generation. In order to test and validate the proposed system, experiments were done in a simulated environment, and results are provided in the thesis to examine the performance of the proposed sketching robot. It is pertinent to note that, as a physical robotic platform was not used for experiments, relevant control algorithms are beyond the scope of this thesis.
    Date of Award2023
    Original languageEnglish
    SupervisorDamith Herath (Supervisor) & Mohammad Abualsheikh (Supervisor)

    Cite this

    '