Skip to content
Open access

METHOD FOR ADAPTIVE COORDINATION OF UNMANNED AERIAL VEHICLE SWARMS BASED ON ARTIFICIAL POTENTIAL FIELDS AND REINFORCEMENT LEARNING

2026 · Scientific papers of Donetsk National Technical University. Series: Informatics, Cybernetics and Computer Science · 0 citations

TL;DR

The developed method provides a mathematically sound architecture for decentralized multi-agent navigation, eliminating the dependency on predefined controller gains while ensuring operational stability.

Abstract

This article presents a method for the adaptive coordination of unmanned aerial vehicle swarms combining artificial potential fields and reinforcement learning. While decentralized swarm navigation frequently relies on artificial potential field algorithms, the use of manually tuned, static interaction parameters limits system performance under variable spatial configurations and local sensing constraints. The proposed method resolves this limitation by formulating dynamic gain selection as a localized optimization process. A Soft Actor-Critic algorithm is used to train a policy that continuously maps localized kinematic observations of nearest neighbors to continuous action outputs, dynamically adjusting the weights for migration, cohesion, and separation. The methodology evaluates multiple reward function configurations, specifically isolating the effects of velocity alignment and movement smoothness, within a three-dimensional point-mass simulation environment. Performance is quantified using spatial separation precision, swarm alignment, total collision rates, and steady-state convergence time. The empirical results demonstrate that a reward formulation based strictly on velocity alignment outperforms both internal variants and an external state-of-the-art fixed-gain controller. The adaptive policy maintains a zero-collision margin, strictly adhering to the physical safety threshold, and achieves steady-state formation convergence significantly faster than static alternatives. The statistical significance of these improvements is confirmed across independent randomized deployments. The developed method provides a mathematically sound architecture for decentralized multi-agent navigation, eliminating the dependency on predefined controller gains while ensuring operational stability.

Read PDF

Similar papers

Open access 2026

Neural state estimation and optimization synthesis of flocking control for mobile robots swarm under uncertainty

The authors propose a method that integrates a learned state estimator based on a fully connected neural network, approximating the particle filter with simultaneous uncertainty evaluation, and an optimization regulator with a control barrier function and a control-Lyapunov function that ensures accuracy comparable to...

S. Kondratev, N. Kazyura, V. Meshcheryakov · 0 citations
Open access Sep 2026

Robust Adaptive Particle Swarm Optimization for High-Level Waypoint Planning in UAV-Swarm Active Source Seeking

Active signal-source seeking by swarms of unmanned aerial vehicles (UAVs) requires reliable waypoint decisions under measurement corruption, misleading response peaks, and geometric constraints. We propose MI-PSO-Adaptive, a high-level waypoint planning framework based on particle swarm optimization (PSO). The framewor...

Yao Meng, Jing Zhou, Yang-Yi Chen et al. · 0 citations
Open access Nov 2026

Safety-Critical Tracking Control of Self-Organizing Swarm for Unmanned Surface Vehicle Based on Identifier–Critic–Actor Reinforcement Learning

In this article, a novel reinforcement learning-based distributed self-organizing swarm optimal control method is presented to address the challenges of real-time collision avoidance and optimal trajectory tracking for multiple unmanned surface vehicles (USVs) operating under uncertain environmental disturbances and in...

Yong-Jin Liu, Xian-Ku Zhang, Hu-Gan Zhang et al. · 0 citations
Preprint Sep 2026

Learning-Based Dynamic Obstacle Avoidance for a UAV Using Only Three Range Sensors

We present a learning-based approach to kinodynamic online motion planning for an Unmanned Aerial Vehicle (UAV) operating at a fixed altitude in unknown dynamic environments, where real-time avoidance of both static and dynamic obstacles must be achieved under conditions of extreme partial observability. The UAV is con...

Mohammad Reza Ranjbar Divkoti, A. Aguiar · 0 citations
Preprint Oct 2026

Learning minimum-time navigation policies in two-dimensional flows with a genetic algorithm

We study minimum-time navigation between two points in a known two-dimensional fluid flow for a vessel with fixed-magnitude slip velocity and controllable direction. The navigation policy is parameterized by a neural network and optimized using a genetic algorithm that evolves an ensemble of candidate strategies throug...

V. Parfenyev · 0 citations

We use cookies to run the site and, with your consent, for analytics and to show ads. See our Cookie Policy.