Professional Documents
Culture Documents
To cite this article: Heather C Humphreys, Young Mi Choi & Wayne J Book (2017): Advanced
patient transfer assist device with intuitive interaction control, Assistive Technology, DOI:
10.1080/10400435.2017.1396564
Download by: [UNIVERSITY OF ADELAIDE LIBRARIES] Date: 25 October 2017, At: 07:16
Advanced Patient Transfer Assist Device with Intuitive Interaction
Control
Heather C. Humphreys1, Young Mi Choi2, Wayne J. Book1
t
ip
Corresponding author:
Downloaded by [UNIVERSITY OF ADELAIDE LIBRARIES] at 07:16 25 October 2017
cr
us
Abstract
This research aims to improve patient transfers by developing a new type of advanced robotic
an
assist device. It has multiple actuated degrees of freedom and a powered steerable base, to
maximize maneuverability around obstacles. An intuitive interface and control strategy allows
M
the caregiver to simply push on the machine in the direction of desired patient motion. The
control integrates measurements of both force and proximity to mitigate any potential large
ed
collision forces and provides operators information about obstacles with a form of haptic
feedback. Electro-hydraulic pump controlled actuation provides high force density for the
pt
actuation.
ce
Nineteen participants performed tests to compare transfer operations (transferring a 250 lb.
Ac
mannequin between a wheelchair, chair, bed and floor) and interaction control of a prototype
device with a commercially available patient lift. The testing included a time study of the transfer
operations and subjective rating of device performance. The results show that operators perform
transfer tasks significantly faster and rate performance higher using the prototype patient transfer
assist device than with a current market patient lift. With further development, features of the
1
new patient lift can help facilitate patient transfers that are safer, easier, and more efficient for
caregivers.
Introduction
t
The goal of this paper is to present data on the effectiveness and efficiency of a newly designed patient
ip
Downloaded by [UNIVERSITY OF ADELAIDE LIBRARIES] at 07:16 25 October 2017
transfer assist device (PTAD) to improve patient transfer tasks (Humphreys, 2016). A new, cost
cr
effective, hydraulically actuated prototype patient transfer assist device has been developed and
fabricated; hydraulic actuation has advantages in terms of force density over electrical actuators that are
us
typically used at this power scale. The PTAD has multiple actuated degrees of freedom, which can
improve maneuverability and reduce forces required from caregivers. Improved methods for control of
an
machines that work collaboratively with humans, sharing a task and a workspace, were developed in this
work. It also aims to overcome some of the control challenges with these actuation systems in this type of
M
application, such as non-ideal characteristics of the low cost actuation systems and management of a
machine with large force capability operating in a home or clinical environment with humans in its
ed
workspace.
pt
Background
ce
This research is motivated out of the growing concern over caregiver injuries related to patient transfers in
the healthcare industry. According to a recent series on National Public Radio titled "Injured Nurses",
Ac
nurses suffer over 35,000 back and other orthopedic injuries that require missed work each year
(Zwerdling 2015). Studies have shown that even the best manual techniques result in unsafe lower back
loads when performing a variety of manual and machine assisted patient transfers (Marras, Davis et al.
1999; Zhuang, Stobbe et al. 1999). Healthcare workers in hospitals, nursing homes, emergency services,
critical care, operating rooms, orthopedic units and home healthcare environments are all potentially
2
exposed to risks associated with patient lifting (Waters, Collins et al. 2006). There is already a need to
find solutions to the current problem, which is likely to increase in the future. For example, as the US
population of adults aged 65 and over grows from 40 million in 2010 to an estimated 55 million in 2020
t
Patient Transfer Barriers
ip
Downloaded by [UNIVERSITY OF ADELAIDE LIBRARIES] at 07:16 25 October 2017
Because of a high incidence of orthopedic injuries to caregivers resulting from manual transfer
cr
operations, many organizations are implementing safe patient handling (SPH) procedures. This includes
organizations such as the Veterans Health Administration (Petzel 2010) as well as guidelines from the
us
Occupational Health and Safety Administration (Chao and Henshaw 2009). Many guidelines restrict
patient lifting without the use of a lift device to reduce the risk of musculoskeletal injuries.
an
Implementation of SPH policies in healthcare settings often leads to significant improvements in
M
musculoskeletal comfort levels as well as reductions in musculoskeletal injuries (Fadul, Brown et al.
2014). There are a number of factors that can create barriers and influence healthcare workers’ motivation
ed
to use lifting devices during patient transfer activities. These include knowledge about existing workplace
guidelines, availability of adequate lifting devices (in both functionality and quantity), adequate facilities
pt
that do not pose barriers (such as too little maneuvering space) to device use, as well as guidance on the
ce
use of a lifting device in an individual patient’s care protocol (Koppelaar, Knibbe et al. 2013).
Ac
A study by (Krill, Staffileno et al. 2012) found similar barriers. They also highlighted the most physically
demanding tasks and environmental limitations expressed by registered nurse staff in their study. The
most demanding tasks included transferring patients from chair to bed, transferring from cart to bed and
handling obese patients in general. The most cited environmental limitations were the size of rooms,
carpeted areas and lack of help. Having mechanical lifts was also cited as an item that would make their
jobs easier.
3
The provision and availability of lift equipment is clearly an important component in reducing caregiver
injuries. It also requires ergonomic assessment and modification of spaces where transfers take place.
Use of floor lifts requires sufficient space around beds and furniture (Collins et al., 2010; Hignett, 2003;
Fray and Hignett, 2013; Koppelaar et al., 2013; Schoenfisch et al., 2011). Factors such as proper lighting
t
must be taken into account (Hingett, Edmonds, etc). The equipment itself must be adequate to the task
ip
Downloaded by [UNIVERSITY OF ADELAIDE LIBRARIES] at 07:16 25 October 2017
and allow caregivers to execute safe patient transfer tasks while keeping physical loads on themselves
cr
within safe levels (Hignett 2003; Collins, Wolf et al. 2004; Kurowski, Gore et al. 2012). Depending on
the task involved, equipment may include ceiling lifts (hoists), floor lifts (and the associated lift slings),
us
standing lifts, slide sheets, electric beds, transfer boards and handling belts.
an
With equipment, environment and policies in place, the reliability of the equipment is important in
reducing injuries in both caregivers and patients. Unreliable equipment will lead operators to lose faith in
M
equipment and stop using them (Elnitsky, Lind et al. 2014). Caregivers may become unwilling to use
battery operated equipment if battery charge runs out in the middle of a lift leaving them without the
ed
ability to complete the task. This can be a problem with ceiling lifts and cause staff to revert to executing
manual lifts. Lack of safety features on equipment can lead to similar issues or injuries to staff or
pt
patients. Equipment that lacks self-locking features can move and lead to falls if the patient moves during
ce
a transfer. A caregiver can more easily be injured when trying to help in these situations.
Ac
Current lifts do not address some needs and do not take advantage of newer developments in robotics
technology. While considerable research has investigated ergonomics of patient transfer operations,
only a few published studies involve development of actual patient transfer assist devices. The original
patent by Hoyer describes a lift device very similar to those on the market today, consisting of a U-shaped
base with casters, a vertical post, and a lifting boom actuated by a hydraulic cylinder (Hildemann and
4
Hoyer 1958). In (Jeannis, Grindle et al. 2013) and (Tsai, Jeannis et al. 2014), a new type of patient lift
called the Strong Arm is developed, which mounts to the base of a power wheelchair. It is also controlled
from an operator input force and uses a sling. The device is powered by electric motors and travels with
the patient on the power chair. However, difficulties with this design include significant lifting power
limitations and tipping stability. In (Mukai, Hirano et al. 2010), design of a humanoid-style nursing
t
assistant robot is developed, which is intended to perform patient lifting and moving tasks. The robot is
ip
Downloaded by [UNIVERSITY OF ADELAIDE LIBRARIES] at 07:16 25 October 2017
led by the caregiver through tactile guidance. Patient comfort is an issue with this design. The National
cr
Institute for Standards and Technology also developed a new type of patient transfer device, which is also
intended for rehabilitation by allowing the patient to walk with part of the body weight supported;
us
however, while this device has a wide range of functionality, it is not a convenient transfer aid
Improving safety in physical human-machine interaction has been receiving increasing attention. Safety
and management of external interaction forces are critical considerations with the proposed design. The
ed
proposed PTAD takes advantage of robotics technology, using coordinated feedback control of powerful
hydraulically actuated joints, which in turn necessitates management of external interaction forces.
pt
Considerable research has also been done to determine safety criteria for robots in operation with human
ce
operators, though no consensus has been reached on suitable standards. Collision forces required to cause
pain or injury vary widely depending on a number of factors, including where the impact occurs on the
Ac
body, the time of impact, and whether or not the body is pinned or free. ISO standards list a maximum
threshold force of 150 N (ISO 2011), but other researchers claim that this threshold is too high (Haddadin,
Albu-Schäffer et al. 2010). In the comprehensive Robotics Handbook (Siciliano and Khatib 2008), a
chapter is devoted to research on safety in physical human-robot interaction, and it includes a function to
rate severity of injury based on impact duration. One of the most common measures of human safety in
robot interaction, the Head Injury Criterion (HIC) (Haddadin, Albu-Schäffer et al. 2010), which is based
5
on acceleration and time of impact. Based on this Park cites a lower maximum collision force target of 50
N. All of the recognized metrics for safety in human robot interaction are noted in literature to have
As for interactions between the machine and the environment, the concept of robot control design for
t
environmental interaction and motion, rather than pure motion control, has been an active area of research
ip
Downloaded by [UNIVERSITY OF ADELAIDE LIBRARIES] at 07:16 25 October 2017
for over three decades, resulting in a substantial volume of publications. Applications of interaction
cr
include human interactions, walking, machining, excavation and teleoperation with haptic feedback. A
considerable amount of research in the 1980's involved hybrid force/position control, or a switching
us
control between position control in free space and force control in contact, e.g. (Raibert and Craig 1981).
A sentinel set of papers was published by Hogan in the late 1980's (Hogan 1985; Hogan 1989), who
an
developed and analyzed an impedance control; in this scheme, rather than controlling force or position,
the controller aims to attain a desired output impedance transfer function Z(s) = F(s)/V (s) , or force divided
M
by velocity. In this case of examining the impedance of a point of mechanical interaction between two
bodies, the machine and the interacting body in the environment, F(s) corresponds to an external
ed
interaction force, and V(s) corresponds to the motion of the interaction point, or how it differs from the
machine speed if there was no interaction force. If there is no external interaction force, then the
pt
Device Requirements
Ac
For this project, the goal was to develop a new patient transfer assist device (PTAD) that is able perform
all typical transfers in a home or clinical environment, work within space constraints of a typical home or
clinical environment, and have adequate lift capacity (at least 500 lb). It also had to be compact and
feature a simple, intuitive control strategy with smooth, stable motion that can allow a single caregiver to
simultaneously maneuver both the device and the patient. A focus group consisting of spinal cord injury
patients, as well as their home caregivers and physical therapists was convened to explore requirements
6
for the new device. The focus group consisted of ten participants, including four spinal cord injury
patients, three home caregivers and three physical therapists. Individual interviews were less structured,
but all included discussion of their experiences with current market patient lifts, limitations of those
devices, and desirable features for new lifts. Individual interviews included home caregivers for a
bariatric paraplegic patient, a hospital nurse, clinicians in a spinal cord injury rehabilitation center,
t
clinicians in a nursing home, sales personnel for a patient lift distributor, and an experienced engineer in
ip
Downloaded by [UNIVERSITY OF ADELAIDE LIBRARIES] at 07:16 25 October 2017
the patient lift industry. The operator testing of the current market patient lift was also performed early in
cr
the development phase. Information from the focus group, individual interviews, and operator testing was
combined to include a broad range of lift users and other stakeholders for the compilation of device
us
requirements. Many of the requirements were also found in existing literature, reinforcing their
importance.
an
Must-Have Machine Features
M
• Capability to perform most typical transfers within a home or institution (e.g. between a
• Provide lift capacity equivalent to current market electrically actuated lifts (500 lb.) in any
• Reach middle of a double bed (beam length 33 in. + 6 in. clearance = 39 in.)
• Legs fit around typical toilets, chairs, recliners (up to 37 in. wide), and bariatric wheelchairs (up
to 35 in. wide)
7
• Compact design to work within the space constraints of a typical home or clinical environment
• Speeds up to 9 in/s
t
ip
• Transfer operations should be completed in less time than with a typical Hoyer lift
Downloaded by [UNIVERSITY OF ADELAIDE LIBRARIES] at 07:16 25 October 2017
cr
Nice-to-Have Machine Features
us
• Capability to perform less common transfers within a home (e.g. if a patient falls in an difficult to
•
pt
• Horizontal motion with respect to the base in two axes, within tipping stability limits
• Ability to control machine with one hand while simultaneously fine-tuning and orienting the
8
patient with the other hand
• Capability to lift, orient and achieve appropriate posture of patient, via caregiver/machine team
t
ip
Downloaded by [UNIVERSITY OF ADELAIDE LIBRARIES] at 07:16 25 October 2017
cr
• Minimal swing and oscillation
us
• Smooth, stable control in all operating conditions
• Controller capability to prevent motion which could cause tipping, to allow for an extended the
an
range of motion
as the starting and ending positions/poses, the distance between locations, and the capabilities/limitations
of the patient, and as a result, there is no industry-defined standard. A comparison of time required,
ed
divided into subtasks, between the new design patient transfer assist device and a current market lift, is
Based on these requirements, a fully functional first generation device was fabricated for testing (Figure
1). It has four actuated degrees of freedom, each powered by its own electro-hydraulic-actuator (EHA).
Ac
All batteries, hydraulic power, actuators, electronics, amplifiers, sensing and controllers are onboard. The
patient rests in a sling that is attached to a hanger bar at the end of the boom; this is the current industry
standard. A force sensing handle is mounted near the patient at the end of the boom to provide all operator
control input. The horizontal boom extension is driven by a belt drive mechanism and moves on a set of
high load capacity linear bearings, and it is powered by a small low speed high torque (LSHT) geroler
9
hydraulic motor. The vertical lifting consists of a scissor mechanism and actuated by a hydraulic cylinder.
The wheels are actuated by larger LSHT hydraulic motors. The design includes obstacle detecting sensors
designed to stop the motion of the lift in the event that an obstacle is encountered.
While this first prototype is considerably bulkier than desired, that extra size was needed for modularity
t
and in order to use currently available mechanical components, in particular the scissor lift mechanism.
ip
Downloaded by [UNIVERSITY OF ADELAIDE LIBRARIES] at 07:16 25 October 2017
Most of the design could easily be optimized to fit in a much more compact package; for example, using
cr
four scissor levels rather than two would considerably decrease the outer dimensions of the scissor
mechanism.
us
The prototype PTAD meets all of the must-have design requirements except for the compact design. The
mechanical design was not optimized for size in this first prototype, and that presented a limitation in
an
testing. The prototype does meet all of the necessary machine design requirements related to power,
speed, lifting capacity, forces required from the operator, battery power, range of motion, and capabilities.
M
It also meets the functional requirements, such as providing smooth, stable, coordinated control of
multiple degrees of freedom and providing the capability for a single operator to control the machine with
ed
one hand, while simultaneously orienting the patient with the other. Some nice-to-have features were not
included in this prototype. Some proposed features that were not implemented in this prototype include:
pt
(1) actuated rotation of the patient between seated and supine posture, (2) compensation for patient
ce
swinging/oscillation, (3) a lightweight, stowable, easily transportable design, (4) payload capacity up to
1000 lb., (5) actuated side-to-side motion, and (6) moveable/spreadable base or outriggers
Ac
A simple, intuitive caregiver interface has been implemented, which provides coordinated rate control
using a force sensing handle mounted near the patient for the operator input. The caregiver interacts
directly with the patient, while simultaneously controlling the lift device. With a powerful machine
working in a relatively delicate environment, it is necessary for the controller to manage any external
10
interaction forces, to keep them in a safe range, in addition to smoothly controlling motion. The operator
An interaction control framework was used to control both the motion of the patient and manage any
external interaction forces, using measurements of force from load cells and proximity from ultrasonic
t
sensors. A significant challenge lies in implementation of interaction control with these electro-hydraulic
ip
Downloaded by [UNIVERSITY OF ADELAIDE LIBRARIES] at 07:16 25 October 2017
pump controlled actuators, which are intrinsically stiff, have slow dynamics, and have many nonlinear or
cr
non-ideal features; the control aspect of the research is discussed in detail in (Humphreys, Book et al.
2013; Humphreys and Book 2014; Humphreys, Book et al. 2014; Humphreys and Book 2016). An
us
impedance control framework has been formulated and implemented, using redundant sensing of
obstacles. In addition to the direct sensing of external force using load cells, an additional feedback term
an
was added to the control using ultrasonic sensors to measure proximity to obstacles. This combination of
proximity and force feedback provides needed redundancy in sensing of obstacles, and the proximity
M
sensing gives the controller some advance notice in order to reduce collision forces by slowing its motion
toward them before they occur. This form of interaction control was initially tested on one degree of
ed
freedom of the machine, as a part of the operator testing. More information on the interaction control,
development and testing can be found in (Humphreys 2016). The addition of robust, reliable interaction
pt
control enables the use of coordinated feedback motion control and an intuitive, easy to use caretaker
ce
Method
Ac
Operator testing of the prototype PTAD was performed and compared against a current market lift: a
Hoyer-style Invacare model 9805 lift. There were two main goals of the operator testing. The first was to
evaluate the performance of the prototype PTAD in patient transfer tasks, focusing on demonstrating the
concept of a multiple degree of freedom force-assist machine. The second was to evaluate the ability of
the controller to manage undesirable external interaction forces. This discussion will focus on the first
11
segment of the operator testing; more information about the interaction control operator testing can be
A total of 19 volunteers participated in the testing of the new PTAD. Each participant independently
performed both the transfer testing and interaction control testing tasks. Once completed, each participant
t
completed a survey designed to capture their opinion of various aspects of the lift such as ability to
ip
Downloaded by [UNIVERSITY OF ADELAIDE LIBRARIES] at 07:16 25 October 2017
complete a transfer alone, controllability of the device and accuracy of the controls. All questions on the
cr
survey were Likert-type items on a 6 point scale, ranging from 1 (very poor) to 6 (excellent). A similar set
of operator experiments was performed using a current market patient lift to transfer the same mannequin,
us
including both the transfer operations and survey, with a different set of 12 participants. Whenever
possible, results are compared between the new PTAD and the current market lift.
an
Participants were recruited through posted flyers on campus and social media, so they were primarily
graduate students at Georgia Tech. No previous experience with patient lifts was required, because such
M
experience could slightly bias the reviews and performance results. Three of the participants did have
minimal prior experience using patient lifts. The experiments with the current market patient lift were
ed
completed earlier, prior to the fabrication of the prototype PTAD, since the results of that study were
intended to not only provide a comparison of the two devices, but to also inform the design of the
pt
prototype PTAD. Just prior to both experiments, participants were given instructions on how to operate
ce
the machines, and they were instructed to perform each type of transfer sub-task (e.g. lift from chair,
lower onto bed, etc.) one time prior to performing the time study.
Ac
All transfer related tests involved moving a 250 lb. mannequin between various locations. These included
1. A wheelchair to a bed
2. A bed to a chair
12
4. From the floor back to a chair.
Transfers to and from the floor were not conducted with the current market lift because it was unable to
reach floor level. The mannequin remained situated in the sling for all operations as the focus was on
investigating the aspects of efficiency and control of the lift while moving the mannequin. The time to
t
complete each component of the transfer task was recorded. For this discussion, two categories of
ip
Downloaded by [UNIVERSITY OF ADELAIDE LIBRARIES] at 07:16 25 October 2017
subtasks are defined. The “transport” subtasks refer to transporting the mannequin from one location to
cr
another, primarily using the base motion; in these stages, the mannequin is free from any physical
interaction with the starting/ending locations (e.g. bed, chair, floor). The “lifting/lowering” subtasks refer
us
to all other subtasks, where the mannequin is in very close proximity to the starting/ending locations.
The experiments were not designed to be identical in all subtasks for both devices, because the devices
an
have somewhat different capabilities to be tested. Some subtasks performed in the experiments (e.g. lift
mannequin from chair, lower mannequin onto bed) were the same for both experiments, and the times can
M
be compared for those. For the transporting stage of the process, it is important for the device to be able to
maneuver the patient through tight spaces with many obstructions. Overall compactness and optimization
ed
of the mechanical design is necessary for that stage, and while that is an important requirement for a
PTAD design, it was not feasible to make the first prototype modular at the same time as optimizing the
pt
size and maneuverability. Therefore, the experiments were designed to avoid any comparison in the
ce
‘transporting’ subtasks. The full transfer operations that were tested are broken down into sets of
subtasks, and only the subtasks that are in boxes in Figures 2-3 can be compared between the two devices.
Ac
The full time study results are valuable for evaluation of the prototype PTAD, with or without the
comparison to the current market device. Images from the testing are shown in Figure 4.
A second set of experiments were performed to evaluate the interaction control, with the same set of
participants as the full transfer experiments with the new PTAD. The experiments focused on the ability
of the controller to manage external interaction forces to remain within safe levels during unexpected
13
collisions and to inform the operator of the collision through a form of haptic feedback. Participants
performed a set of short transfers of the mannequin between two chairs. In some cases, an obstacle was
placed such that it was not visible to the operator, and the collision forces were measured; this was tested
with the interaction control active and inactive. This comparison of the collision forces and motion
tracking performance between the cases of active and inactive interaction control provided an evaluation
t
of the controller’s ability to reduce the interaction forces and inform the caretaker of the obstacle.
ip
Downloaded by [UNIVERSITY OF ADELAIDE LIBRARIES] at 07:16 25 October 2017
Each participant completed a total of eight transfers. They completed two transfers for practice, one with
cr
the obstacle and one with no obstacle. Then they completed six transfers, each with different
combinations of three control algorithm variations and two obstacle variations (with or without the
us
obstacle). Half of the tests included the obstacle. The interaction control was tested using only the
horizontal and vertical motions of the machine; these results can be easily extended to the wheel motion.
an
It is preferable to test the interaction control on the horizontal motion, to avoid complications from
gravitational forces and complex kinematics. The test obstacle consisted of a small steel plate (3 in. x 6
M
in.) and a stiff spring, which were mounted on the device, and it was hidden from the view of the
operator. The details of the interaction control design and subsequent experimental evaluation are a
ed
separate thrust of this research and beyond the scope of this publication. Further information on the
interaction control can be found in (Humphreys 2016; Humphreys and Book 2016).
pt
The time taken to perform the subtasks of the transfer operations with the market lift and the new PTAD
are shown in Figures 5 and 6. These are all standard box plots, with the whiskers representing the
Ac
minimum to maximum values, the lower box representing the first quartile, the upper box representing the
third quartile, and the line between the boxes representing the median. The time study gives some insight
into which operations require the most time, and subsequently how design efforts should be focused to
14
It is promising to note that most transfer subtasks require considerably less time with the new PTAD..
And surprisingly the data show that the “transport" subtasks are generally not slower for a motions of
similar distance, even with this oversized prototype. With the current market lift, most lifting and
lowering operations are typically performed in about 40-50 seconds, while with the PTAD, those same
subtasks are typically performed in about 15-20 seconds. With some improvements to optimize the
t
mechanical design to make the gross motions around the room more efficient, a PTAD can be expected to
ip
Downloaded by [UNIVERSITY OF ADELAIDE LIBRARIES] at 07:16 25 October 2017
cr
Unpaired t-tests were used to compare the times required for participants to perform the subtasks that
could be directly compared between the experiments using the current market patient lift and the new
us
PTAD. All data were checked to ensure that all assumptions were met for the test (independence, normal
distribution and equality of variances for ANOVA). All passed with the exception of the case of 'fine
an
position over bed' as described below:
• Lift above wheelchair: There was a significant difference in the time required, with the new
M
PTAD (M=52.62, SD=17.86) requiring an average of 36.15 seconds less than the current market
• Fine position over bed: There was a significant difference in the time required, with the new
PTAD (M=15.58, SD=4.77) requiring an average of 38.76 seconds less than the current market
pt
There was a single outlier in the market lift data (a duration more than 4 times the average of all
other) making the distribution non-normal. Removing this point satisfies the normality
Ac
requirement and the difference is still significant [the PTAD requiring an average of 27.82
seconds less than the current market lift (M=43.4, SD=18.85); t(27)=6.2, p<<0.001.]
• Lower onto bed: There was a significant difference in the time required, with the new PTAD
(M=17.63, SD=3.1) requiring an average of 23.63 seconds less than the current market lift
15
• Lift above bed: There was a significant difference in the time required, with the new PTAD
(M=20.05, SD=6.43) requiring an average of 32.05 seconds less than the current market lift
• Lower onto chair: There was a significant difference in the time required, with the new PTAD
(M=13.79, SD=5.79) requiring an average of 39.26 seconds less than the current market lift (M=53.05,
t
SD=21.66); t(28)=7.53, p<<0.001.
ip
Downloaded by [UNIVERSITY OF ADELAIDE LIBRARIES] at 07:16 25 October 2017
cr
For the interaction control experiments, descriptive statistics were calculated across all trials, as well as a
two-way ANOVA to compare the difference in means of the external interaction forces between cases
us
with the interaction control active and inactive. The results show that the difference in means is highly
significant (p << 0.001), has a medium effect size (0.525), and a mean reduction in collision force of
an
53%. Furthermore, in all cases with interaction control active, collision forces were maintained below 140
N, which is below the ISO standard maximum acceptable collision force of 150 N. Considerable analysis
M
of the control design and details on the interaction control experiments can be found in (Humphreys
2016). These interaction control results indicate that it should be feasible to design a hydraulically
ed
actuated patient transfer assist device that has control capability to manage any unintentional collision
The survey results (Figures 7-9) showed that overall, participants rated the PTAD higher than the current
market lift in almost every metric, though the differences were small in some cases. In the overall ratings,
Ac
the average ratings for important metrics such as the ability to transfer alone, ease of obstacle avoidance,
ability to transfer quickly, and overall maneuverability, all were improved over the current market lift, in
spite of the bulky prototype design. For the PTAD controller ratings, all of the metrics including overall
controllability, accuracy, response speed and smoothness of the lifting arm, accuracy and response speed
of the base, all had average ratings of good to excellent. The only exception was the smoothness of the
base motion, which was rated as fair. Average ratings for all metrics were higher than those for current
16
market lift except the speed of response of the lifting arm.
Operators found the smoothness of the base control to be particularly lacking; it is difficult to attain
smooth motion control performance with these low cost power units, but it could likely be improved with
better compensation for the non-ideal effects. Operators also found the speed of response of the lifting to
t
be lower than desired, which is partially a result of integrating the operator input to obtain a position
ip
Downloaded by [UNIVERSITY OF ADELAIDE LIBRARIES] at 07:16 25 October 2017
control reference, as well as the need for considerable damping in the control to obtain fairly smooth
cr
motion from these power units. This effect could likely be improved to some degree with better tuning.
Several operators also noted that the machine dynamic response was not what they initially hoped for, but
us
they felt that they could “get used to the machine dynamics" quickly. Ratings were divided into “lifting
and lowering" stage and a “transfer stage", where the transfer stage refers to the motions involved in
an
transporting the mannequin from one place to another, when it is free from any contact with the starting
The ratings for the transfer stage are primarily focused on the wheel control and base design, both of
ed
which are not yet optimized in this prototype. Still, the ratings showed improvement over the current
market device in all metrics, and all metrics received average ratings between good and excellent, except
pt
for “ease of changing directions", which is primarily a result of the casters. A number of operators
ce
mentioned the known limitations, such as the difficult to turn casters and large overall size. But they also
noted that the powered wheels are a significant improvement over the unpowered casters of the current
Ac
market lift. Smooth control of fine motions is difficult with these hydraulic power units, as a result of the
check valve circuit designed to hold the load against gravity without power. The check valves are pilot
operated and remain closed until the pump reaches sufficient pressure to open them; that combined with
17
The most positive ratings and review comments were given for the lifting and lowering
stages (Figure 10), with average ratings for all metrics falling between very good and excellent, and
improvement in average ratings over the current market lift for all four metrics.
Conclusion
A prototype PTAD is developed in this research, and it demonstrates the concepts of multiple electro-
t
ip
hydraulically actuated degrees of freedom with coordinated control, as well as interaction control, applied
Downloaded by [UNIVERSITY OF ADELAIDE LIBRARIES] at 07:16 25 October 2017
to the patient transfer application. Some issues with patient lifts are directly improved with this prototype:
cr
• Lower forces required from the operator
us
• Powered base motion
performance improvements. First, the addition of multiple hydraulically actuated degrees of freedom with
ed
intuitive coordinated control can be expected to lead to the following, in later stages, after an optimized
Second, the addition of intuitive coordinated control, where the operator simply pushes on the machine in
Ac
the desired direction of patient motion, can be expected to lead to reduced operator mental workload.
Third, the addition of interaction control can be expected to result in the following improvements, once it
18
• Fewer incidences of damage to objects in the workspace.
Addition of more actuated degrees of freedom with coordinated, feedback controlled motion has
considerable advantages in this application. But without the addition of interaction control, the fly-by-
t
wire, feedback controlled system would have potential to increase the possibility of injuries or damage
ip
Downloaded by [UNIVERSITY OF ADELAIDE LIBRARIES] at 07:16 25 October 2017
resulting from collisions. In the current market lift, the actuated joint is controlled by an operator in open
cr
loop from a manual hydraulic input lever; in such systems, the operator can feel any significant increases
in the force applied by the actuator. Adding feedback motion control can improve the machine dynamic
us
response and enable coordinated control, but it reduces the operator's feel for the environment. Under
pure position or velocity feedback, when the machine encounters an obstacle, the feedback control
an
increases the actuator command in an attempt to achieve the desired motion; therefore, as compared with
open loop manual hydraulic control, the operator is less aware of the increase in forces applied to the
M
environment. Furthermore, the addition of more actuated degrees of freedom can result in higher collision
force capabilities from some joints (e.g. the wheels, which are not actuated in the current market patient
ed
lift). The addition of various forms of interaction control typically provide the operator with even better
haptic information about interaction forces than open loop control, and it allows for the controller to
pt
reduce the interaction forces. Additionally, like driving a car with ultrasonic backup sensors, if the
ce
machine can be trusted to warn the caregiver of any impending external interactions, and to minimize
interaction forces, then the caregiver will be able to spend less time checking around the machine to avoid
Ac
collisions while moving through the workspace, thereby reducing the caregiver's mental workload and
19
References
Bostelman, R., J. Albus, et al. (2007). Recent developments of the HLPR Chair. 2007 IEEE 10th
International Conference on Rehabilitation Robotics, IEEE.
Chao, E. L. and J. L. Henshaw (2009). Guidelines for Nursing Homes Ergonomics for the
Prevention of Musculoskeletal Disorders. U. D. o. L. P. N. O. Occupational Safety and
Health Administration.
Collins, J. W., L. Wolf, et al. (2004). "An evaluation of a ‘‘best practices’’ musculoskeletal injury
t
prevention program in nursing homes." Injury Prevention 10(4): 206-211.
ip
Downloaded by [UNIVERSITY OF ADELAIDE LIBRARIES] at 07:16 25 October 2017
Elnitsky, C. A., J. D. Lind, et al. (2014). "Implications for patient safety in the use of safe patient
handling equipment: A national survey " International journal of nursing studies 51(12):
cr
1624-1633.
Fadul, R. M., L. M. Brown, et al. (2014). "Improving transfer task practices used with air
travelers with mobility impairments: A systematic literature review." Journal of public
us
health policy 35(1): 26-42.
Haddadin, S., A. Albu-Schäffer, et al. (2010). Safe physical human-robot interaction:
measurements, analysis and new insights. Robotics research, Springer Berlin Heidelberg:
395-407.
an
Hignett, S. (2003). "Intervention strategies to reduce musculoskeletal injuries associated with
handling patients: a systematic review." Occup. Environ. Med. 60(9): E6.
Hildemann, V. R. and T. R. Hoyer (1958). Adjustable base invalid lift. U.S. Patent No. 2, 406.
M
Hogan, N. (1985). "Impedance control: An approach to manipulation." Journal of Dynamic
Systems and Control 107(8).
Hogan, N. (1989). Controlling impedance at the man/machine interface. IEEE International
Conference on IEEE.
ed
Humphreys, H. and W. J. Book (2014). Advanced hydraulically actuated patient transfer assist
device. 8th FPNI Ph.D. Symposium on Fluid Power, American Society of Mechanical
Engineers.
ce
Humphreys, H. C., W. J. Book, et al. (2013). Hydraulically actuated patient transfer device with
passivity based control. ASME/BATH 2013 Symposium on Fluid Power and Motion
Control,, American Society of Mechanical Engineers.
Humphreys, H. C., W. J. Book, et al. (2014). Caretaker-machine collaborative manipulation with
an advanced hydraulically actuated patient transfer assist device. ASME 2014 Dynamic
Systems and Control Conference, American Society of Mechanical Engineers.
ISO (2011). DIS 13482: Robots and robotic devices–Safety requirements–Non-medical personal
care robot. Geneva, Switzerland.
20
Jeannis, H., G. G. Grindle, et al. (2013). Initial development of direct interaction for a transfer
robotic arm system for caregivers. In Rehabilitation Robotics (ICORR), 2013 IEEE
International Conference on.
Koppelaar, E., J. J. Knibbe, et al. (2013). "The influence of individual and organisational factors
on nurses’ behaviour to use lifting devices in healthcare." Applied Ergonomics 44(4):
532-537.
Krill, C., B. A. Staffileno, et al. (2012). "Empowering staff nurses to use research to change
practice for safe patient handling." Nursing Outlook 60(3): 157-162.
t
Kurowski, A. S., R. P. Gore, et al. (2012). "Differences among nursing homes in outcomes of a
ip
safe resident handling program." J. Healthc. Risk Manag. 32(1): 35-51.
Downloaded by [UNIVERSITY OF ADELAIDE LIBRARIES] at 07:16 25 October 2017
Marras, W. S., K. G. Davis, et al. (1999). "A comprehensive analysis of low-back disorder risk and
spinal loading during the transferring and repositioning of patients using different
cr
techniques." Applied Ergonomics 42(7): 904-926.
Mukai, T., S. Hirano, et al. (2010). Development of a nursing-care assistant robot RIBA that can
us
lift a human in its arms. Intelligent Robots and Systems (IROS), 2010 IEEE/RSJ
International Conference on, IEEE.
Petzel, R. (2010). VHA Directive 2010-032: safe patient handling program and facility design. W.
D. Department of Veterans Affairs.
an
Raibert, M. H. and J. J. Craig (1981). "Hybrid position/force control of manipulators." Journal of
Dynamic Systems, Measurement, and Control 103(2): 126-133.
Siciliano, B. and O. Khatib (2008). Springer handbook of robotics, Springer Science & Business
M
Media.
Tsai, C. Y., H. Jeannis, et al. (2014). "Stability analysis of electrical powered wheelchair-mounted
robotic-assisted transfer device." Journal of rehabilitation research and development
51(5): 761.
ed
Waters, T., J. Collins, et al. (2006). "NIOSH research efforts to prevent musculoskeletal disorders
in the healthcare industry." Orthopaedic Nursing 25(6): 380-389.
Zhuang, Z., T. J. Stobbe, et al. (1999). "Biomechanical evaluation of assistive devices for
transferring residents." Applied Ergonomics 30(4): 285-294.
pt
Zwerdling, D. (2015). NPR special series: Injured nurses - hospitals fail to protect nursing staff
from becoming patients.
ce
Ac
21
Downloaded by [UNIVERSITY OF ADELAIDE LIBRARIES] at 07:16 25 October 2017
Ac
ce
pt
ed
M
an
us
Figuure 1: Four deegree of freeddom prototyppe patient traansfer assist ddevice.
cr
ip
22
t
Downloaded by [UNIVERSITY OF ADELAIDE LIBRARIES] at 07:16 25 October 2017
Ac
ce
pt
ed
M
an
Figure 2.. Operator innput force maapping to patiient motion
us
cr
ip
23
t
t
ip
Downloaded by [UNIVERSITY OF ADELAIDE LIBRARIES] at 07:16 25 October 2017
cr
Figure 3. Trannsfer locationns; Left: Currrent market lift, Right: Paatient Transfe
F fer Assist Devvice
us
an
M
ed
pt
ce
Ac
24
Downloaded by [UNIVERSITY OF ADELAIDE LIBRARIES] at 07:16 25 October 2017
Ac
ce
pt
ed
M
an Figuure 4. Photos from transfeer operations testing
us
cr
ip
25
t
t
ip
Downloaded by [UNIVERSITY OF ADELAIDE LIBRARIES] at 07:16 25 October 2017
cr
us
an
Figure 5. Time sttudy results for the PTA
AD transfer tasks. Boxees indicate subtasks
s thaat can be
M
directly compared bbetween the current marrket lift andd the prototyype PTAD.
ed
pt
ce
Ac
26
t
ip
Downloaded by [UNIVERSITY OF ADELAIDE LIBRARIES] at 07:16 25 October 2017
cr
us
Figure 66. Time studyy results for tthe market lifft transfer tassks. Boxes inndicate subtasks that can bbe directly
coompared betw
ween the currrent market llift and the pprototype PTA
AD.
an
M
ed
pt
ce
Ac
27
Downloaded by [UNIVERSITY OF ADELAIDE LIBRARIES] at 07:16 25 October 2017
Ac
ce
Figure 7. O
pt
ed
M
an Overall lift raatings. Left: PTAD, Righht: market lifft
us
cr
ip
28
t
Downloaded by [UNIVERSITY OF ADELAIDE LIBRARIES] at 07:16 25 October 2017
Ac
ce
pt
ed
Figure 8. Liift controllerr ratings. Left
M
an
ft: PTAD, Rigght: market liift
us
cr
ip
29
t
Downloaded by [UNIVERSITY OF ADELAIDE LIBRARIES] at 07:16 25 October 2017
Ac
ce
pt
ed
M
an
us
Figure 9. Lift transfer stagge ratings. Leeft: PTAD, Right: markett lift
cr
ip
30
t
t
ip
Downloaded by [UNIVERSITY OF ADELAIDE LIBRARIES] at 07:16 25 October 2017
cr
us
Fiigure 10. Ratings for thee lifting and lowering stagge. Left: PTA
AD, Right: cuurrent markeet lift
an
M
ed
pt
ce
Ac
31