UAV Path Planning using Global and Local Map Information with Deep Reinforcement Learning

  title={UAV Path Planning using Global and Local Map Information with Deep Reinforcement Learning},
  author={Mirco Theile and Harald Bayerlein and Richard Nai and David Gesbert and Marco Caccamo},
  journal={2021 20th International Conference on Advanced Robotics (ICAR)},
Path planning methods for autonomous unmanned aerial vehicles (UAVs) are typically designed for one specific type of mission. This work presents a method for autonomous UAV path planning based on deep reinforcement learning (DRL) that can be applied to a wide range of mission scenarios. Specifically, we compare coverage path planning (CPP), where the UAV's goal is to survey an area of interest to data harvesting (DH), where the UAV collects data from distributed Internet of Things (IoT) sensor… 

Figures and Tables from this paper

Multi-UAV Path Planning for Wireless Data Harvesting With Deep Reinforcement Learning
This work formulate the path planning problem for a cooperative, non-communicating, and homogeneous team of UAVs tasked with maximizing collected data from distributed IoT sensor nodes subject to flying time and collision avoidance constraints, and solves the problem through a deep reinforcement learning (DRL) approach.
Long Endurance Flight Testing Results for the UIUC-TUM Solar Flyer
This paper describes the long-endurance flight testing results of the UIUC-TUM Solar Flyer from the Summer and Fall of 2020, and describes on-board data processing for real-time mission adaptation.
Multi-Agent Belief Sharing through Autonomous Hierarchical Multi-Level Clustering
This work proposes autonomous hierarchical multi-level clustering (MLC), which forms a clustering hierarchy utilizing decentralized methods and is used as a backbone to solve the communication problem for locally-interactive applications such as UAV tracking problems.


UAV Path Planning for Wireless Data Harvesting: A Deep Reinforcement Learning Approach
This work proposes a new end-to-end reinforcement learning approach to UAV-enabled data collection from Internet of Things (IoT) devices in an urban environment that enables the agent to make movement decisions for a variety of scenario parameters that balance the data collection goal with flight time efficiency and safety constraints.
UAV Coverage Path Planning under Varying Power Constraints using Deep Reinforcement Learning
This work proposes a new method to control an unmanned aerial vehicle (UAV) carrying a camera on a CPP mission with random start positions and multiple options for landing positions in an environment containing no-fly zones.
Distributed and Energy-Efficient Mobile Crowdsensing with Charging Stations by Deep Reinforcement Learning
This paper proposes a distributed control framework for energy-efficient and DIstributed VEhicle navigation with chaRging sTations, called “e-Divert”, a distributed multi-agent deep reinforcement learning (DRL) solution, which uses a convolutional neural network to extract useful spatial features as the input to the actor-critic network to produce a real-time action.
Coordinated Control of UAVs for Human-Centered Active Sensing of Wildfires
A dual-criterion objective function based on Kalman uncertainty residual propagation and weighted multi-agent consensus protocol is developed, which enables the UAVs to actively infer the wildfire dynamics and parameters, track and monitor the fire transition, and safely manage human firefighters on the ground using acquired information.
Accessing From the Sky: A Tutorial on UAV Communications for 5G and Beyond
This article gives a tutorial overview of the recent advances in UAV communications, with an emphasis on how to integrate UAVs into the forthcoming fifth-generation (5G) and future cellular networks.
An Integrated Traveling Salesman and Coverage Path Planning Problem for Unmanned Aircraft Systems
A preliminary study on how to formulate and solve the TSP-CPP problem, where two novel approaches including a grid-based approach and a dynamic programming based approach are introduced to find the (near) optimal solution.
Decentralized Multi-Agent Deep Reinforcement Learning in Swarms of Drones for Flood Monitoring
This work studies the use of Deep Q-Networks (DQN) as the optimization strategy for the trajectory planning that is required for monitoring floods, and trains agents over simulated floods in procedurally generated terrain and demonstrates good performance with two different reward schemes.
Design Challenges of Multi-UAV Systems in Cyber-Physical Applications: A Comprehensive Survey and Future Directions
This survey aims to pinpoint the most fundamental and important design challenges of multi-UAV systems for CPS applications, highlighting key and versatile aspects that span the coverage and tracking of targets and infrastructure objects, energy-efficient navigation, and image analysis using machine learning for fine-grained CPS applications.
Distributed Wildfire Surveillance with Autonomous Aircraft using Deep Reinforcement Learning
Teams of autonomous unmanned aircraft can be used to monitor wildfires, enabling firefighters to make informed decisions. However, controlling multiple autonomous fixed-wing aircraft to maximize fo...