Skip to main content

LUP Student Papers

LUND UNIVERSITY LIBRARIES

Tether-free Driveline Control for Water Propulsion Devices

Marklund, Isak LU and Müller, Axel LU (2023) MAMM01 20232
Ergonomics and Aerosol Technology
Certec - Rehabilitation Engineering and Design
Abstract
For many machines, safety requires the constant presence of an operator, with
the risk of damage or danger if the operator is unintentionally absent. Dead
man’s switches (DMS) are commonly used to halt operations if this absence is
detected, often relying on physical elements like leashes or buttons. However,
these solutions can hinder operator convenience and equipment aesthetics. For
devices with electrical brushless DC motors, motor data can be generated and
analyzed. This opens the door for machine learning models to discern patterns
signifying operator presence. The potential benefits are most evident in personal
devices, where user activity directly influences motor data. This study delves into
the feasibility of a... (More)
For many machines, safety requires the constant presence of an operator, with
the risk of damage or danger if the operator is unintentionally absent. Dead
man’s switches (DMS) are commonly used to halt operations if this absence is
detected, often relying on physical elements like leashes or buttons. However,
these solutions can hinder operator convenience and equipment aesthetics. For
devices with electrical brushless DC motors, motor data can be generated and
analyzed. This opens the door for machine learning models to discern patterns
signifying operator presence. The potential benefits are most evident in personal
devices, where user activity directly influences motor data. This study delves into
the feasibility of a software-based, tether-free, DMS for a jet surfboard.
Initially, user opinions on the existing leash-and-magnet DMS were gathered,
assessing its impact on user experience through surveys and interviews. Results
showed that a tether-free DMS solution would improve user experience mainly
in regards to convenience and comfort. The existing DMS is regarded as simple,
safe and reliable - characteristics that would need to be reflected by a tether-free
DMS.
Subsequently, machine learning models were constructed to analyze ride log
data in order to detect surfer presence. Two approaches were explored, detecting
when someone is falling off the board, and detecting when the board is running
with no one on it. It was found that strong negative and positive trends in the
motor data (corresponding to deceleration and acceleration) were the characteristics that the models mostly identified with fall-off and off-board data, which
resulted in some misclassifications. Results show that there are some possible
differentiating features between different activities on board in some cases, but
the models ultimately fall short of the performance requirements for a DMS. (Less)
Popular Abstract
Many devices require a constant human presence for safe
operation, often utilizing dead man’s switches (DMS) to inter-
vene and disable the device when an operator is absent. These
conventional safety mechanisms, such as leashes or buttons,
can sometimes compromise user convenience and equipment
design. By conducting a user study, the demand for a tether-free
DMS was evaluated, investigating the DMS on a jet-surfboard
and how it impacts the user experience. In smaller personal
devices equipped with brushless DC motors, like jet-propelled
surfboards, an opportunity arises by enabling the evaluation of
motor data in real-time. Through the use of machine learning,
this study aimed to identify patterns in the data that could
... (More)
Many devices require a constant human presence for safe
operation, often utilizing dead man’s switches (DMS) to inter-
vene and disable the device when an operator is absent. These
conventional safety mechanisms, such as leashes or buttons,
can sometimes compromise user convenience and equipment
design. By conducting a user study, the demand for a tether-free
DMS was evaluated, investigating the DMS on a jet-surfboard
and how it impacts the user experience. In smaller personal
devices equipped with brushless DC motors, like jet-propelled
surfboards, an opportunity arises by enabling the evaluation of
motor data in real-time. Through the use of machine learning,
this study aimed to identify patterns in the data that could
indicate the absence of an operator. (Less)
Please use this url to cite or link to this publication:
author
Marklund, Isak LU and Müller, Axel LU
supervisor
organization
course
MAMM01 20232
year
type
H2 - Master's Degree (Two Years)
subject
keywords
Dead man’s switch, electric surfboard, jet surfboard, machine learning, user experience, usability testing, neural network, decision tree, bootstrap aggregating, time series
language
English
id
9141654
date added to LUP
2023-11-29 13:31:01
date last changed
2023-11-29 13:31:01
@misc{9141654,
  abstract     = {{For many machines, safety requires the constant presence of an operator, with
the risk of damage or danger if the operator is unintentionally absent. Dead
man’s switches (DMS) are commonly used to halt operations if this absence is
detected, often relying on physical elements like leashes or buttons. However,
these solutions can hinder operator convenience and equipment aesthetics. For
devices with electrical brushless DC motors, motor data can be generated and
analyzed. This opens the door for machine learning models to discern patterns
signifying operator presence. The potential benefits are most evident in personal
devices, where user activity directly influences motor data. This study delves into
the feasibility of a software-based, tether-free, DMS for a jet surfboard.
Initially, user opinions on the existing leash-and-magnet DMS were gathered,
assessing its impact on user experience through surveys and interviews. Results
showed that a tether-free DMS solution would improve user experience mainly
in regards to convenience and comfort. The existing DMS is regarded as simple,
safe and reliable - characteristics that would need to be reflected by a tether-free
DMS.
Subsequently, machine learning models were constructed to analyze ride log
data in order to detect surfer presence. Two approaches were explored, detecting
when someone is falling off the board, and detecting when the board is running
with no one on it. It was found that strong negative and positive trends in the
motor data (corresponding to deceleration and acceleration) were the characteristics that the models mostly identified with fall-off and off-board data, which
resulted in some misclassifications. Results show that there are some possible
differentiating features between different activities on board in some cases, but
the models ultimately fall short of the performance requirements for a DMS.}},
  author       = {{Marklund, Isak and Müller, Axel}},
  language     = {{eng}},
  note         = {{Student Paper}},
  title        = {{Tether-free Driveline Control for Water Propulsion Devices}},
  year         = {{2023}},
}