Page 17 - Preliminary Analysis Towards a Standardized Readiness Framework - Interim Report

Preliminary Analysis Towards a Standardized Readiness Framework

such as ad hoc networking. The agents may be integrated with models such as reinforcement
learning and route optimization algorithms.

Use case provided by Istanbul Technical University and Turkcell that aims to harness the
advancements in reinforcement learning (RL) to enhance the deployment, route selection,
and coordination of unmanned aerial vehicles (UAV) in disaster scenarios [52], especially for
scenarios that require immediate response such as earthquakes and floods. This case study
emphasizes its use of ad hoc networks among drones.

Enhancing the efficiency of response efforts increases resilience and accelerates recovery in
communities affected by disasters. Delays, resource limitations, and logistical challenges often
hamper traditional disaster response efforts. To overcome these obstacles, a coordinated
UAV network is designed to autonomously perform essential tasks within disaster-stricken
areas. Utilizing RL algorithms, UAVs can learn and adjust their operations (including route
navigation, returning to charging stations, and data detection and transmission) based on
feedback from the environment. In particular, the project integrates several state-of-the-art RL
approaches, such as multiagent learning (for achieving efficient cooperation among UAVs),
sim2real transfer (for leveraging simulated data), and curriculum learning (for achieving a
smoother learning curve from simple to complex scenarios). This combination of approaches
allows for the optimization of task distribution and resource management in real-time, while
ensuring generalization across a rich variety of disaster scenarios.

UAVs will be equipped with sensors, cameras, communication systems, and payload delivery
mechanisms. The drones collaborate to carry out a range of tasks such as reconnaissance,
damage assessment, communication relay, and aid distribution. Through advanced data
collection and mapping algorithms, the UAV network achieves real-time situational awareness,
facilitating informed decision-making by the response teams.

To achieve the tasks, each UAV maintains a connection to ground stations, either through direct
links or an ad-hoc network, ensuring seamless coordination and data exchange.
The Ad-hoc networks used in the use case greatly facilitate communication among drones due
to their advantage in flexibility, mobility, cost-effectiveness, resilience, and scalability.

Since a disaster might destroy the existing infrastructure, the flexibility provided by ad-hoc
networks makes them ideal for temporary or emergencies where establishing traditional
networks is impossible. Devices in an ad-hoc network can move freely without affecting the
network's overall functionality, which makes the use of ad-hoc networks ideal for disaster
detection. Since ad-hoc networks don't require dedicated infrastructure, they can be more
cost-effective to deploy. Ad-hoc networks can also be more resilient compared to traditional
networks, as they don't rely on a central point of control. Ad-hoc networks can easily scale
to accommodate many devices, making it easy to scale to a large number of drones in real
applications.

9