A Framework for Testing and Evaluation of Operational Performance of Multi-UAV Systems
| Autor: | Abdollah Homaifar, Xuyang Yan, Kyriakos G. Vamvoudakis, Bruce J. Holmes, Shamila Nateghi, Mrinmoy Sarkar |
|---|---|
| Rok vydání: | 2021 |
| Předmět: | |
| Zdroj: | Lecture Notes in Networks and Systems ISBN: 9783030821920 IntelliSys (1) |
| DOI: | 10.1007/978-3-030-82193-7_24 |
| Popis: | In this paper, we propose a data-driven testing and evaluation framework for multi-UAVs to evaluate their performance in executing missions in the physical world. Seven micro-behaviors, termed here as modes of operation, are leveraged to describe the autonomous functionalities of the UAVs. These functionalities are then used to design five scenarios for model training, validation and testing of the proposed framework. Each scenario includes a distinct sequence of behaviors for the UAVs in order for the different autonomous functionalities to be evaluated. We develop and implement a simulation environment using the Robot Operating System (ROS), Gazebo, and the Pixhawk autopilot to generate synthetic data for the training of a classification model. This trained model is then utilized to evaluate the behaviors of the UAVs while performing real-world missions. Finally, the proposed framework is tested using synthetic data generated from a simulation environment and validated using real-world data. |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|