Fair path planning for Unmanned Aerial Vehicles (UAVs) in emergency response missions
The development of autonomous systems, such as Unmanned Aerial Vehicles (UAVs), has witnessed significant growth in recent years. Their deployment in emergency response missions has become increasingly crucial due to their potential benefits, such as faster response times. However, these benefits al...
- Autores:
-
Pardo González, Germán Roberto
- Tipo de recurso:
- Trabajo de grado de pregrado
- Fecha de publicación:
- 2023
- Institución:
- Universidad de los Andes
- Repositorio:
- Séneca: repositorio Uniandes
- Idioma:
- eng
- OAI Identifier:
- oai:repositorio.uniandes.edu.co:1992/67649
- Acceso en línea:
- http://hdl.handle.net/1992/67649
- Palabra clave:
- Column Generation
Q-Learning
Reinforcement Learning
OR in emergency response
Drone Deployment
Multi-objective optimisation
Ingeniería
- Rights
- openAccess
- License
- http://creativecommons.org/licenses/by-nd/4.0/
| Summary: | The development of autonomous systems, such as Unmanned Aerial Vehicles (UAVs), has witnessed significant growth in recent years. Their deployment in emergency response missions has become increasingly crucial due to their potential benefits, such as faster response times. However, these benefits also raise ethical considerations that must be addressed when utilizing them in emergency scenarios. This work aims to contribute to the fair automation of UAVs in disaster response missions. To achieve this, the Column Generation method is employed to solve the initial routing problem within a specific grid. The primary objectives of this approach are twofold: to maximise population coverage and maximise fairness in the distribution of coverage, with a focus on assisting the most vulnerable individuals. Since the data accounts for two conflicting objectives, only Pareto-optimal solutions are considered and compared based on the assigned weights for each objective. Moreover, the Q-Learning algorithm is utilised to dynamically evaluate and prioritise newly discovered survivors who were not initially expected. This algorithm enables the agent to adapt and make decisions based on the vulnerability of the survivors. In order to address uncertainties inherent in emergency situations, a random variable is incorporated, enabling the agent to learn and make decisions based on new information. |
|---|