Skip to main content

LUP Student Papers

LUND UNIVERSITY LIBRARIES

Audio-based Motion Detection

Gomatam, Gopal LU and Dujmic, Filip (2021) EITM02 20211
Department of Electrical and Information Technology
Abstract
Motion detection is an essential technology and it has numerous use-cases, such as security tracking, automated door opening systems, and IP cameras. Commonly, passive infrared sensors or radio frequency sensors are used for motion detection. However, this thesis focuses on performing motion detection using existing hardware at audio frequencies. Specifically, it aims to implement a motion detection algorithm using existing speakers and microphones on each of two intercom devices designed by Axis Communications, AXIS I8016-LVE Network Video Intercom (Device 1) and AXIS A8207-VE Mk II Network Video Door Station (Device 2). The goal of implementing such an algorithm is to save power and cost by reusing existing hardware.

In this thesis,... (More)
Motion detection is an essential technology and it has numerous use-cases, such as security tracking, automated door opening systems, and IP cameras. Commonly, passive infrared sensors or radio frequency sensors are used for motion detection. However, this thesis focuses on performing motion detection using existing hardware at audio frequencies. Specifically, it aims to implement a motion detection algorithm using existing speakers and microphones on each of two intercom devices designed by Axis Communications, AXIS I8016-LVE Network Video Intercom (Device 1) and AXIS A8207-VE Mk II Network Video Door Station (Device 2). The goal of implementing such an algorithm is to save power and cost by reusing existing hardware.

In this thesis, data was collected and studied for both stationary and moving objects to understand the systems’ behaviour. The range detection capability for different signal parameters was compared and analysed using a stationary object, at different positions within the range of the systems. This was done to find the optimum signal parameters for the motion detection algorithm. Once these were determined, data was collected and analysed for a moving object. Based on this data, an optimisation and tracking algorithm was designed to obtain better results from the systems. The results show that for Device 1, the algorithm can detect any movement of a person within the range of the system. However, for Device 2, the algorithm can differentiate between different movements, allowing the detection of a person who is approaching the device, rather than just someone who is just passing by. The latter enables power saving for the device since its camera will only turn on when someone approaches the device with the intention of using it. (Less)
Popular Abstract
Over the past few decades, motion detection has played a vital role in automating various tasks. These tasks fall into the domains of security monitoring, automated doors and lighting control, among others. Common sensors for motion detection include those that emit and receive electromagnetic waves (RF) as well as sensors that just detect electromagnetic (infrared) radiation emitted from the human body. However, when these sensors are not readily available, other technologies can be modified to achieve the same purpose. In this thesis, one such example of reusing existing speaker and microphones to perform motion detection at audio frequencies was studied. The modification was implemented on two intercom devices, AXIS I8016-LVE Network... (More)
Over the past few decades, motion detection has played a vital role in automating various tasks. These tasks fall into the domains of security monitoring, automated doors and lighting control, among others. Common sensors for motion detection include those that emit and receive electromagnetic waves (RF) as well as sensors that just detect electromagnetic (infrared) radiation emitted from the human body. However, when these sensors are not readily available, other technologies can be modified to achieve the same purpose. In this thesis, one such example of reusing existing speaker and microphones to perform motion detection at audio frequencies was studied. The modification was implemented on two intercom devices, AXIS I8016-LVE Network Video Intercom (Device 1) and AXIS A8207-VE Mk II Network Video Door Station (Device 2), built by Axis Communications. The main goal of implementing motion detection capability in the intercom devices is to enable power saving by turning on the camera in the device only when a person is detected to approach the device rather than just walking by. Moreover, the use of existing hardware for this purpose allows for significant cost saving.

The first step in the implementation was to study the behaviour of the systems for a stationary object. This was carried out by collecting data for different object locations with different signal parameters within the effective ranges of both devices. The collected data provided the mean and standard deviation of the error between the measured distance and the actual distance from the device to the object. The results were analyzed to determine the optimum duration, type, shape and frequency range of the signal for obtaining good distance estimation. The next step was to employ the chosen parameters to study the systems' behaviour under different scenarios for a moving object. Data was collected for five different scenarios with the aim to investigate whether both devices can differentiate between different types of movements in front of the device. Based on this data, an optimization and tracking algorithm was implemented to obtain more accurate results from both systems.

The results show that although both devices have a mean error in the range of $10$ cm for a stationary object, they can still detect motion within their effective ranges. Device 1 cannot differentiate between the different scenarios considered and it will always turn on when there is any movement within its range. However, Device 2 can differentiate between different scenarios considered, hence allowing it to determine if a person is approaching the device or just passing by. This means that it is more efficient in saving power and the associated cost. This is in contrast to using passive infrared sensors, as these recognise even small movements within their effective range, and they cannot differentiate between different kinds of movements, i.e. the device turns on even when it is not required. (Less)
Please use this url to cite or link to this publication:
author
Gomatam, Gopal LU and Dujmic, Filip
supervisor
organization
course
EITM02 20211
year
type
H2 - Master's Degree (Two Years)
subject
keywords
sonar, motion detection, audio frequencies, audio processing
report number
LU/LTH-EIT 2021-825
language
English
id
9055280
date added to LUP
2021-06-16 14:10:48
date last changed
2021-09-28 10:28:53
@misc{9055280,
  abstract     = {{Motion detection is an essential technology and it has numerous use-cases, such as security tracking, automated door opening systems, and IP cameras. Commonly, passive infrared sensors or radio frequency sensors are used for motion detection. However, this thesis focuses on performing motion detection using existing hardware at audio frequencies. Specifically, it aims to implement a motion detection algorithm using existing speakers and microphones on each of two intercom devices designed by Axis Communications, AXIS I8016-LVE Network Video Intercom (Device 1) and AXIS A8207-VE Mk II Network Video Door Station (Device 2). The goal of implementing such an algorithm is to save power and cost by reusing existing hardware.

In this thesis, data was collected and studied for both stationary and moving objects to understand the systems’ behaviour. The range detection capability for different signal parameters was compared and analysed using a stationary object, at different positions within the range of the systems. This was done to find the optimum signal parameters for the motion detection algorithm. Once these were determined, data was collected and analysed for a moving object. Based on this data, an optimisation and tracking algorithm was designed to obtain better results from the systems. The results show that for Device 1, the algorithm can detect any movement of a person within the range of the system. However, for Device 2, the algorithm can differentiate between different movements, allowing the detection of a person who is approaching the device, rather than just someone who is just passing by. The latter enables power saving for the device since its camera will only turn on when someone approaches the device with the intention of using it.}},
  author       = {{Gomatam, Gopal and Dujmic, Filip}},
  language     = {{eng}},
  note         = {{Student Paper}},
  title        = {{Audio-based Motion Detection}},
  year         = {{2021}},
}