Cloud4AuRoRA: an open and interactive framework for UAV simulation

• Ano de defesa: 2022
• Autor(a) principal: Fagundes Junior, Leonardo Alves
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: eng
• Instituição de defesa: Universidade Federal de Viçosa; Ciência da Computação
• Palavras-chave em Português: Computação em nuvem - Métodos de simulação; Robótica; Drone; Google Colaboratory; Software - Desenvolvimento; Aprendizado do computador; Ciência da Computação
• Link de acesso: https://locus.ufv.br//handle/123456789/30492; https://doi.org/10.47328/ufvbbt.2022.538

To verify solutions before implementing them in a robot, robots and developers need an experimentation platform that accurately reproduces the real-world environment as well as the robot’s physical interactions with that environment. Through simulation, they can evaluate the robot’s performance in terms of localization, planning, and motion control. Over the past ten years, most of the progress in robotics developed in academia has been made using well-known software such as MATLAB/Simulink, Robot Operating System (ROS), Copelia Sim, and Gazebo. However, learning to use and program these platforms can be challenging for students. Nonetheless, the current body of literature does not provide comprehensive frameworks to adequately address this issue in conjunction with learning about modeling (kinematics and dynamics) and controlling the unmanned aerial vehicle (UAV) to perform different types of missions. This dissertation presents a UAV simulation tool for learning robotics on Google Cloud Platform for developing control strategies, parameter tuning, and evaluation on different types of tasks such as positioning, trajectory planning, trajectory tracking, and robot cooperation. The proposed framework, called cloud4AuRoRA, can also contribute to advances in scientiĄc research in aerial robotics, speciĄcally UAV control and navigation, and machine learning applications. Following the framework of the AuRoRA platform, developed in MATLAB, the approach proposed here merges robot modeling and control using C/C++, and data storage and display using Python. The proposed framework is intended to facilitate the development and validation of studies in aerial robotics for students and researchers by improving results and proof-of-concept development time, aided by the high computational performance provided by Google Collaboratory. Cloud4AuRoRA showed a 70 times performance improvement compared to MATLAB when running the same routine, through direct conversion/translation between programming languages. Finally, it is worth mentioning that could4AuRoRA is adaptable for implementation in a GPU environment through modiĄcations to the core code of the platform. Keywords: Aerial Robotics. Simulation on the Cloud. Google Colaboratory. Modeling and Control. Learning Technology. Student Assessment.