A Bioinspired Neural Network-Based Approach for Cooperative Coverage Planning of UAVs

Simone Godio,Stefano Primatesta,Giorgio Guglieri,Fabio Dovis

doi:10.3390/info12020051

Simone Godio, Stefano Primatesta + Show 2 more

Open Access

https://doi.org/10.3390/info12020051

Copy DOI

Journal: Information	Publication Date: Jan 25, 2021
Citations: 13	License type: CC BY 4.0

Affiliation: Polytechnic University of Turin

Abstract

This paper describes a bioinspired neural-network-based approach to solve a coverage planning problem for a fleet of unmanned aerial vehicles exploring critical areas. The main goal is to fully cover the map, maintaining a uniform distribution of the fleet on the map, and avoiding collisions between vehicles and other obstacles. This specific task is suitable for surveillance applications, where the uniform distribution of the fleet in the map permits them to reach any position on the map as fast as possible in emergency scenarios. To solve this problem, a bioinspired neural network structure is adopted. Specifically, the neural network consists of a grid of neurons, where each neuron has a local cost and has a local connection only with neighbor neurons. The cost of each neuron influences the cost of its neighbors, generating an attractive contribution to unvisited neurons. We introduce several controls and precautions to minimize the risk of collisions and optimize coverage planning. Then, preliminary simulations are performed in different scenarios by testing the algorithm in four maps and with fleets consisting of 3 to 10 vehicles. Results confirm the ability of the proposed approach to manage and coordinate the fleet providing the full coverage of the map in every tested scenario, avoiding collisions between vehicles, and uniformly distributing the fleet on the map.

Highlights

Autonomous exploration with mobile robots is a widespread problem in robotics [1].Even if this topic has been widely studied since the last decade [2], there are still some open problems, including coverage planning [3]
The PX4 autopilot is used to control unmanned aerial vehicles (UAVs) simulated with Gazebo using the Software In The Loop (SITL) framework [33], where SITL allows PX4 to be executed without using any hardware
We present an innovative approach to solve the coverage planning problem with a fleet of UAVs

Summary

Introduction

Autonomous exploration with mobile robots is a widespread problem in robotics [1]. Even if this topic has been widely studied since the last decade [2], there are still some open problems, including coverage planning [3]. An interesting approach was proposed in [17] introducing a preliminary model for fleet coordination in urban environments considering a distribution of docking stations Another approach proposed in [18] solves the coverage planning problem by defining a series of waypoints for UAVs to explore maps. This work aims to solve a coverage planning problem to explore and monitor a specific area with a fleet of UAVs. the goal is to cover the entire map using a fleet of UAVs maintaining, at the same time, a uniform distribution of the fleet on the map, as well as avoiding collision between vehicles and other obstacles. An initial configuration with UAVs already uniformly distributed in the map would have some benefits on the performance of the coverage planning since it is an optimal starting condition obtaining a full coverage of the map in less time with fewer moves

Proposed Approach

Preliminary Simulations

Realistic Simulations

Findings

Conclusions and Further Developments