DAMIANO BRUNORI

• (Deep) Reinforcement Learning ((D)RL)

• Multi-agent systems and related learning/execution paradigms:

  • 1. fully centralized with a base station acting as a main supervisor/controller

    2. fully decentralized

    3. centralized training and decentralized execution mainly based on an actor-critic neural network structure

    4. shared parameter training

• Delayed Markovian Decision Processes with Mixed Observability (D-MOMDP) solved through (D)RL:

  • 1. observation delays

    2. action delays

• Path Planning (mainly combined with (D)RL)

(Deep) Reinforcement Learning

• Multiple Unmanned Aerial Vehicle (Multi-UAV) systems:

  • 1. homogeneous systems (e.g., only quadcopters UAVs)

    2. cooperation

    3. area coverage

    4. communication maximization through smart services provision

    5. desired target spotting 

    6. desired target tracking

Optimization algorithms (e.g., Symplex), Supervised Learning, Deep Learning

• Multiple Unmanned Aerial Systems (Multi-UASs) in the U-Space:

  • 1. heterogeneous systems (e.g., fixed wings and quadcopters UAVs)

    2. trajectory conflicts analysis

    3. trajectory conflicts estimation

    4. deployment optimization based on specific separation minima requirements

    5. flight level assignment