Microservices for autonomous UAV inspection with UAV simulation as a service

Autonomous UAV systems are increasingly touted as the dominant future paradigm for inspecting civil infrastructure. Within this study, we have designed and developed a cloud system for high-level path planning, monitoring, and testing of autonomous UAV missions for inspecting infrastructures such as power lines, power towers, bridges, and railways. The software architecture is based on identified system's functional and non-functional requirements. The cloud system is intended for UAV inspection operators and therefore should support multiple concurrent users. The microservice architecture has assured independence between functionalities, allowing independent scaling resulting in the fast processing time of near optimal route calculation for UAVs reaching inspection targets. Furthermore, the independence between the services facilitates feature addition and future development. The system robustness is assured by containerizing services and continuously deploying to the Kubernetes cluster distributed across multiple worker nodes. Kubernetes scaling properties have enabled multiple concurrent users regardless of heavy computations for inspection paths. Application load testing has resulted in low processing time when individual services are scaled. Calculated inspection paths are validated for real-world inspection by employing the UAV Gazebo simulation based on a 3D dynamic model with an onboard flight controller, leading the UAV through the waypoints provided by the cloud system. By containerizing the simulation and deploying it within the cluster, we have enabled developers and users to test paths before sending the real-world UAVs to the inspection.

Automated summary

This study presents the design and development of a cloud-based system for advanced path planning, monitoring, and testing of autonomous UAV missions in inspecting various infrastructures. The system relies on a microservice architecture to ensure independence and fast processing. By containerizing and continuously deploying services, the system achieves robustness and supports multiple concurrent users.