hero-image
HOME
hero-image

UAV Swarm Acoustic Simulation Tool (MATLAB)

hero-image
Abhinaya Parameswaran

Project Timeline

Aug 2021 - May-2022

OVERVIEW

Designed and lead a MATLAB simulation project to model and optimize the acoustic behavior of UAV swarms in urban environments, for UC Berkeley's MEng Master's Capstone. The system combines discrete element and ray-based dynamic models to simulate vehicle motion and sound propagation, using machine learning to fine-tune kinematic parameters for minimal overall noise. The simulator employs a Forward Euler solver to update drone positions, velocities, accelerations, and orientations in real time based on kinematic differential equations. It generates 3D noise maps using an inverse-radius attenuation model to estimate sound levels in decibels (dB). A custom 3D urban environment was also developed—including skyscrapers, residential buildings, and road networks—each defined with obstacle meshes for autonomous path planning and collision avoidance. The simulation successfully demonstrated how different building materials affect sound absorption, with interior noise levels measured between 50–56 dB, providing a foundation for quieter swarm operation strategies in complex environments.

HighlightS

  • Integrated machine learning to minimize swarm noise through optimized kinematic parameters.
  • Created 3D urban environments with obstacle-aware drone navigation and noise mapping.
  • Demonstrated realistic interior noise levels (50–56 dB) across materials with varying absorption.

SKILLS

MATLAB Simulation and ModelingControl System DesignMulti-Agent (Swarm) Behavior AnalysisAcoustic Propagation and MappingMachine Learning (Applied Opitmization)Real-time numerical integration (Forward Euler method)Data VisualizationMultidisciplinary Problem Solving

SUPPORTING MATERIALS

Additional Details

Problem Statement

With the dramatic rise of commercial drones, noise production is becoming a growing concern in urban, residential, rural, and natural environments. We are building a simulator which models the propagation and absorption of acoustical energy in the surrounding environment as opposed to massive physical experiments with real resources and testing opportunities. Using MATLAB as the programming tool, we modeled the physical characteristics of noise production and figured out its influence on the environment for the use of companies such as Amazon or GrubHub who would benefit from simulation trials before live implementation of package delivery.

Drone and Noise Mapping Prototype

https://youtu.be/w1mBzYm4A2s

In the video above, we can see 4 different scenarios of a flock of 4 drones. Each scenario models a different type of motion that the four drones could make. The first scenario involved the four drones diverging in the x-y plane at the same height. The second scenario involved the flock of drones traveling in a wave horizontally in the direction the two leading drones are present. The third scenario involved all four drones traveling up vertically at fixed distances from each other’s side. The fourth scenario involves drones diverging from each other but also rising at the same time. The motion in each scenario was coded by designing/coding the kinematic equations in MatLab and updating the position through time discretization and displaying them with Matlab’s built in frame capture functionality to later save them as avi files.

Developing the Urban Environment

The main goal of designing this effective urban environment was to have diversity of buildings, a realistic city layout, and have an environment relevant to the intended value of our simulation tool. To model a delivery service environment drones are initialized at the left hand side of the map in the blue industrial buildings as their base. They are then expected to use the forward euler algorithm to find the most efficient path to the residential targets on the right hand side of the map to delivery packages at the purple residential buildings. We included a wide avenue in the middle to model an almost mainstreet like avenue that cuts through the commercial district and industrial region to give a clear path for drones to travel to their targets.


Since the building materials are marked with different colors in the urban map environment, sound waves for the sound could be blocked by the buildings, so we designed four different view angles for the urban environment. Also, with different view angles, we could have a better understanding of the noise propagation throughout the whole area. Moreover, we are basically setting the yellow buildings as commercial or factory buildings that are made of concrete, the transparent ones would be the skyscrapers, and the purple ones would be the residential apartment and houses made of bricks. Each building material would have their own unique absorption coefficient that influences the noise levels generated by drones nearby.


With multiple drones deployed from the left hand side of the urban map, drones would automatically avoid the obstacles in front of them and then fly across this city district. After applying the absorption effect from the surrounding buildings, the noise level would be reduced to below 50 decibels for people who are working inside the buildings. As for the noise on the street, the maximum noise level would be reduced from more than 80 decibels to below 60 decibels. The tool simulates the drones flying across a certain downtown area with both commercial and residential buildings presented, and with the absorption aspect implemented, it would successfully simulate the noise effect to different groups of people.



lowinertia
Portfolio Builder for Engineers
Created by Aram Lee
© 2025 Low Inertia. All rights reserved.