The Network and Analytics (NAN) group participates in Drones4Safety on behalf of Aarhus University. It is part of the Department of Electrical and Computer Engineering and as a group focuses on applying Internet technology to real world and large-scale applications.
In the Drones4Safety project, we primarily focus on the collaborative multi-drone nature of the project. In other words, our work ensures that a fleet of drones collaborates by making joint decisions and execute missions in a safe fashion: for examples coordinating and synchronizing the drones’ movement by using a leader-follower scheme, coordinating path planning to optimize the inspection path around a piece of infrastructure and optimizing the schedule of battery charging.
ENHANCING THE BATTERY CAPACITY OF THE DRONES: the challenge
With current state-of-the-art, the battery capacity is not sufficient to execute a full mission at once, and therefore drones are expected to need to recharge their batteries during the mission at one (of the) charging station(s), a fixed location nearby the infrastructure. We are working on a charging scheduler that finds a mathematically optimal charging schedule such that drones are never with an empty battery, and that two drones do not charge simultaneously at a charging station. By taking into account the path that each drone is taking in the near future, such an optimal schedule ensures that the overall mission execution time is minimized. Furthermore, we are working on implementing our theoretical work in practise to show the feasibility in a practical setting and to discover limitations with our found methods. This involves both simulations of mission executions and testing with real-world and controlled environments.
COLLABORATION AND AUTONOMY: the two key aspects of the mission
Our work focuses on increasing both the collaborative nature and the autonomy of the project. These two aspects have the potential of making bridge and railway inspections faster and more robust system. Using a single drone can be risky, because any failure of the drone means that the inspection cannot be continued at all. In a fleet of drones in which decisions are made autonomously and dynamically, the fleet can adapt to a single drone falling and still finish the mission.
LIMITS OF AUTOMATION: unforeseen events and risks
The use of collaborative and autonomous drones can lead to execute inspections without the (or with minimal) assistance of a human operator. It is therefore highly important that drones can operate safely not only in optimal environment, but far-from-ideal environments as well. Persons are pretty good at handling unforeseen situations, but this “intelligence” must be carried over to drones too. As an example, a drone – whose battery unexpectedly fully depletes due to weather conditions or unexpected obstacles – must always be to land safely to prevent any potential accidents. As a research project, we cannot take into account every single problem that can arise, but at least we should have a sense of what can go wrong in any step of the inspection mission. Bringing the results of D4S to a commercial solution will take significant extra effort to provide strong safety guarantees.
DRONES4SAFETY: a step forward to prevent infrastructure failures
I hope that our work eventually leads to high-quality and frequent inspection of all transport infrastructure in Europe. This is the best step forward to identify and prevent future infrastructure failures. Furthermore, I wish that the research output of this project can also inspire and be a basis for future work on different drone-related research fields.
Kaspar David Hageman, Postdoc Researcher at the Network and Analytics group (Aarhus University).