Loading...

Cooperative Search and Localization by Using Multiple UAVs based on Reinforcement Learning

Zahmatkesh, Mohsen | 2021

159 Viewed
  1. Type of Document: M.Sc. Thesis
  2. Language: Farsi
  3. Document No: 54165 (45)
  4. University: Sharif University of Technology
  5. Department: Aerospace Engineering
  6. Advisor(s): Saghafi, Fariborz
  7. Abstract:
  8. Nowadays, the use of drones in search, rescue, delivery, internet sharing, and even entertainment operations is massively expanding. Therefore, receiving data from environment in order to identify the access routes and other variables can help us to save time. As requirements to achieve this goal are development of intelligent algorithms that allow complex tasks to be performed automatically.One of these algorithms is the reinforcement learning, which enables drones to plan routes needed to perform the above operations even in conditions where the environment is dynamic or there is no precise model. Intelligent cooperation between drones through data exchange can considerably improve the level of success. In addition, if there is an operational problem for an intelligent agent, other agents will be able to continue the mission. By exchanging data, internal and external disturbances and system uncertainties can be reduced and eliminated.In this research, the problem of search and rescue operations by a group of quadrotors using the reinforcement learning algorithm and adaptive sliding mode controller is investigated. One of the common problems of using reinforcement learning algorithms in continuous environments is increasing dimensions of the value-action matrix, which increases the amount of memory occupied and in turn reduces the convergence speed of the algorithm. In the present study, the convergence speed of the algorithm is increased by discretization of the surrounding environment. To be close to real situation, it is assumed that the surrounding environment is unknown and has different obstacles. The effect of the number of drones on the convergence time of the algorithm has also been investigated
  9. Keywords:
  10. Reinforcement Learning ; Neural Networks ; Sliding Mode Control ; Cooperative Search and Localization ; Cooperative Intelligence

 Digital Object List

 Bookmark

No TOC