Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

On rapid parallel tuning of controllers of a swarm of MAVs -- distribution strategies of the updated gains

Dariusz Horla, Wojciech Giernacki, Vít Krátký, Petr Štibinger, Tomáš Báča, Martin Saska

TL;DR

This work tackles the challenge of rapidly tuning MAV swarm controllers without a model by introducing a deterministic, model-free parallel tuning framework that scales from a pair to a swarm. It combines a master/slave, zero-order equal-division search (EQL) with either averaging across drones or distributed configurations to balance reliability and tuning speed. The method is validated in simulation and real-world flights, showing that averaging improves repeatability under noise and disturbances, while distributed configurations can shorten the tuning horizon at the cost of longer or more variable runtimes. Practically, this approach enables faster, more reliable deployment of MAV swarms across different payloads and environments, with potential to establish a parameter bank for look-up-table style gains.

Abstract

In this paper, we present a reliable, scalable, time deterministic, model-free procedure to tune swarms of Micro Aerial Vehicles (MAVs) using basic sensory data. Two approaches to taking advantage of parallel tuning are presented. First, the tuning with averaging of the results on the basis of performance indices reported from the swarm with identical gains to decrease the negative effect of the noise in the measurements. Second, the tuning with parallel testing of varying set of gains across the swarm to reduce the tuning time. The presented methods were evaluated both in simulation and real-world experiments. The achieved results show the ability of the proposed approach to improve the results of the tuning while decreasing the tuning time, ensuring at the same time a reliable tuning mechanism.

On rapid parallel tuning of controllers of a swarm of MAVs -- distribution strategies of the updated gains

TL;DR

This work tackles the challenge of rapidly tuning MAV swarm controllers without a model by introducing a deterministic, model-free parallel tuning framework that scales from a pair to a swarm. It combines a master/slave, zero-order equal-division search (EQL) with either averaging across drones or distributed configurations to balance reliability and tuning speed. The method is validated in simulation and real-world flights, showing that averaging improves repeatability under noise and disturbances, while distributed configurations can shorten the tuning horizon at the cost of longer or more variable runtimes. Practically, this approach enables faster, more reliable deployment of MAV swarms across different payloads and environments, with potential to establish a parameter bank for look-up-table style gains.

Abstract

In this paper, we present a reliable, scalable, time deterministic, model-free procedure to tune swarms of Micro Aerial Vehicles (MAVs) using basic sensory data. Two approaches to taking advantage of parallel tuning are presented. First, the tuning with averaging of the results on the basis of performance indices reported from the swarm with identical gains to decrease the negative effect of the noise in the measurements. Second, the tuning with parallel testing of varying set of gains across the swarm to reduce the tuning time. The presented methods were evaluated both in simulation and real-world experiments. The achieved results show the ability of the proposed approach to improve the results of the tuning while decreasing the tuning time, ensuring at the same time a reliable tuning mechanism.
Paper Structure (5 sections, 4 equations, 7 figures)

This paper contains 5 sections, 4 equations, 7 figures.

Table of Contents

  1. Introduction
  2. Experimental hardware platform & controller
  3. Tuning procedure
  4. Experimental results
  5. Summary

Figures (7)

  • Figure 1: Hardware platforms (a pair of MAVs)
  • Figure 2: Surface of the logarithm of the sum of absolute tracking errors in Z averaged over 20 experiments
  • Figure 3: Gazebo simulation (RVIZ) presenting circular trajectory with the ideal/implemented thrust force vector (left) and 2 MAV models in the simulation (left)
  • Figure 4: Surface of logarithm of the sum of absolute tracking errors in Z averaged over 20 experiments, with a payload
  • Figure 5: Circular reference trajectories
  • ...and 2 more figures