Qualitative perspectives from Kinova^® Jaco^® robotic arm owners: understanding daily device usage

Abstract

Purpose:

Although the benefits of Assistive Robotic Manipulators (ARMs) have been recognized since their emergence over the past two decades, the number of ARM owners remains limited, though expected to grow. This study interviews current owners to gather their perspectives and uses the Human Activity Assistive Technology (HAAT) model to understand their usage experiences with ARMs.

Materials and methods:

A semi-structured remote interview study was conducted with robotic arm owners (≥3 months). Pre-interview surveys collected demographics and common tasks with the robotic arm. Thematic analysis identified key themes.

Results:

Eleven ARM owners were recruited and gave examples of the Kinova^® Jaco^® robotic arm being a life-changing benefit promoting independence which emerged as a theme. Additional themes emerged around operation, caregiver influence, environment, and training. Common tasks included drinking and opening doors. Design recommendations focused on modular attachments, adjustable speed, smoother operation, intuitive control, and enhanced safety.

Conclusion:

We performed a qualitative interview and drew conclusions guided by the HAAT model to better understand interactions between the user, the robotic arm, and contexts of its use for manipulation tasks. The insights gained can better inform prospective owners and guide future research in assistive robotics.

1. Introduction

Previous works that discuss the usability of assistive robotic arms often do so through objective outcomes which lack an understanding of the implications of a person’s personal lived everyday robotic arm use. By framing this work using the Human Activity Assistive Technology (HAAT) model, we aim to categorize and understand the everyday user’s lived assistive technology experience [1].

Robotic arms, initially designed for industrial and surgical purposes, have evolved to include wheelchair-mounted robotic manipulators [2], offering greater options for individuals with upper limb and mobility impairments. These devices offer increased independence in performing daily activities, reducing their reliance on caregivers, and improving their social participation and overall quality of life [3–6].

In the 1990s, several wheelchair-mounted robotic arms, including Manus and iARM developed by Exact Dynamics, or the first robotic arm approved by the United States Food and Drug Administration (FDA), the Raptor (Applied Resource Corporation, approved in 2002) [7,8], were available. However, currently, the Kinova^® Jaco^® robotic arm is the sole FDA and International Organization for Standardization (ISO)-approved wheelchair-mounted robotic arm distributed in the United States. The first commercial Kinova assistive robotic arm, the Jaco [9], was introduced in 2009. According to a Kinova representative, “there are nearly 200 Jaco users in the U.S., and 600 worldwide, and that number is growing quickly as awareness of the benefits of assistive robots increases” (April 2023).

The increasing availability of US insurance support and the growing adoption of Jacos worldwide are expected to lead to a rise in the number of robotic arm owners. Therefore, it’s essential to understand the experiences of current owners and how they integrate the robotic arm into their daily lives. By comprehending how the robotic arm is presently used from direct owner feedback, identifying areas for enhancement, and understanding its integration into users’ daily routines, future iterations can align with user requirements, ultimately promoting wider adoption of these devices.

A substantial portion of research in the field of robotics involves novice users who receive only brief training sessions, often less than an hour in duration, in controlled lab environments. While such studies serve to establish initial efficacy [10], or analyze task performance metrics [11], they have limitations associated with lab settings. These limitations include less training time, inadequate device acclimation, limited opportunity for having the robotic arm mounted on their wheelchair, and device operation where users are not required to get to the arm menu through their wheelchair menu which is nontraditional operation. These studies often guide future design efforts by focusing on reducing cognitive load but may not address realistic multi-step tasks in real environments.

In contrast to these earlier studies, highlighting the need for algorithmic advancement to address cognitive challenges in manual control, owners typically master basic robotic arm control modes within a few months. However, a comprehensive understanding of the task context, environment, and the user’s operational abilities is necessary for characterizing end user robotic arm usage preferences. Current owners which continuously practice with the 24/7-mounted robotic arm on their wheelchairs can provide insights into these contexts. Unlike short-duration lab studies, only a few recent works have begun to emphasize the importance of understanding the long-term impact and benefits of robotic arm use [4,6,12].

Previous research with experienced robotic arm owners is limited. Early studies, such as Rommer’s exploration of MANUS arm usage over 12 weeks, concluded that long-term arm usage offers economic and quality of life benefits, including increased employment for ARM users [13]. Another study in 2013 interviewed iARM owners in the Netherlands, highlighting improved quality of life and frequent use for tasks like eating grasp-able foods, drinking, and picking up objects [14]. While these studies touch on long-term benefits, their focus is primarily on arm tasks and usage.

A more in-depth examination of the long-term impact of Jaco comes from Beaudoin et al. who followed seven long-term Jaco owners over 6 months [12]. This study acknowledges that past research often concentrates on short-term impacts with new users and Beaudoin assesses various measures, including life habits, user satisfaction, and psychosocial aspects through validated questionnaires. They concluded the Jaco increases user participation in life habits. However, this work lacks qualitative perspectives on controlling the robotic arm.

In contrast, our qualitative work conducts semi-structured interviews to gain deeper insights into arm acquisition, training, and the integration of Jaco into users’ lives. A recent study also takes a qualitative approach [15]. Atigossou and colleagues recognize the importance of stakeholder perspectives to explain complex interaction between perceived usage benefits and nuanced human preferences. Their study focuses more on social implications related to human assistance and cost, while we explore user habits, caregiver relationships, and control preferences, providing a more comprehensive understanding of the user experience.

2. Objective

The study’s objective is to understand the perceptions of current robotic arm owners and explore the daily operational benefits and concerns related to the Jaco robotic arm. The HAAT model is a framework [16] for understanding a more holistic combination of the effective interaction between the user and assistive device during manipulation tasks through varying contexts. Therefore, a contribution of this study is to use the HAAT framework (Figure 1) to examine qualitative perspectives of robotic arm use, focusing on how assistive technology (AT) factors impact independence and could contribute to a positive quality of life for current users.

Figure 1.

The Human Activity Assistive Technology (HAAAT) model from Assistive technologies: Principles practice, 5th ed., cook, am, etc.

The HAAT model consists of four interrelated components: Human, which includes the physical, cognitive, and sensory capabilities of the person; Activity, which refers to the specific task goal, in this case manipulation tasks performed using the robotic arm; Assistive Technology, represented by the robotic arm; and Context, the environment where the activity takes place, including social, cultural, and physical factors.

In the discussion section, we will analyze how human, activity, and contextual factors influence the effectiveness and usability of the robotic arm. While the robotic arm (AT) remains constant, the nuanced interaction between users (Human), tasks (Activity), and environment (Context) reveal important considerations for its improvement.

3. Methods

We conducted a 90-min one-on-one semi-structured interview with a total of eleven individuals who currently own robotic arms. These interviews were conducted remotely over Zoom and recorded with participants’ consent.

After obtaining verbal consent, each participant was requested to complete two surveys through RedCap before their scheduled interview date. These surveys included a demographic questionnaire and a self-reported upper limb assessment, detailed in Data collection and analysis section. On their designated interview date, each participant was presented with a detailed interview guide using a shared slideshow. All three investigators convened to establish the interview guide, which comprised larger research questions and follow-up questions (see Appendix A). At the conclusion of the interview, participants were sent a gift card. This manuscript covers sections of a larger interview, but omits control modality questions covered in previous publications [17,18].

This study was conducted through the University of Pittsburgh School of Health and Rehabilitation Sciences and Technology, IRB approved Pittpro number STUDY22020152.

3.1. Participant recruitment

Owners were recruited by distributing flyers to lists of current robotic arm users. This included reaching out to a US Kinova representative, a US distributor (Kinova Jaco distributor), and the Jaco user Facebook group. The flyer instructed potential subjects to contact a research coordinator for additional information. In response to the flyer, interested individuals directly contacted the research team. Subsequently, they underwent a telephone screening to determine study eligibility.

3.1.1. Participant selection

Participants qualified if they were 18 years of age or older, used a power wheelchair as their primary means of mobility, had an upper limb impairment preventing completion of daily manipulation tasks, and had owned the arm for at least three months. Participants unable to sit for the full interview duration were excluded from the study.

3.2. Data collection and analysis

While this is primarily a qualitative study, before the semi-structured interview, we collected data through two surveys: a demographic questionnaire and a functional self-assessment via the Capabilities of Upper Extremity Questionnaire [19]. The results of these surveys are presented separately from the qualitative themes, but analysis from both the surveys and interview are combined when interpreting results in the discussion section.

The demographic questionnaire captures information about age, gender, diagnosis, self-described skill level, and duration of robotic arm ownership. Users also indicated what control input they used for robotic arm operation and how their robotic arm was mounted. The questionnaire also included a Likert scale for assessing general attitudes and preferences toward technology. In the last section of the survey participants entered three common tasks they use the robotic arm for, and also ranked the difficulty of performing fourteen tasks with the robotic arm which were pre-listed in a table.

The Capabilities of Upper Extremity Questionnaire (CUE-Q) is a 32-item patient-reported self-assessment that evaluates arm, hand, and bilateral function, including reaching, pulling, turning their hand, grasping, and pinching. It has demonstrated validity and reliability in adult populations [20]. Each item is scored on a five-point scale, with 0 indicating unable to complete and 4 indicating no difficulty. Scores range from 0 to 128, with higher scores indicating better upper limb function.

For interview analysis, first interviews were de-identified and transcribed using the NVivo qualitative analysis software (Lumivero, Denver, CO), followed by manual corrections to the transcripts which further familiarized investigators with the data. Inductive thematic analysis was used to develop an understanding of shared experiences from robotic arm owners [21]. Subsequently, two investigators independently identified initial categories based on interview questions but also added notes about the broader interview takeaways. Then this baseline led to inductively finding patterns in the broader transcripts toward abstracted concepts. It took three meetings to establish a Codebook. The first meeting grouped repetitive patterns where investigators met to establish codes toward a first representative Codebook. Applying the initial Codebook, one randomly selected interview was coded by two coders (the first two authors). A second meeting was then used to discuss and further refine the Codebook, after which a randomly selected second interview was chosen to use the refined Codebook for initial coverage metrics. Using NVivo, coverage scores were generated after which the third meeting occurred to discuss excerpts highlighted for disagreement and review coverage and kappa scores. Following this, they coded two additional randomly selected interviews, resulting in improved kappa scores of 0.77 and 0.83, with an average score of 0.80 for the two interviews. The remaining interviews were then coded by one researcher based on this finalized Codebook but continued to be refined through team feedback until theme saturation was reached resulting in the eight themes presented in Thematic findings section.

4. Results

We separately present results on survey findings and qualitative themes. In our survey findings we group information into human and activity characteristics by presenting participant characteristics and then robotic arm usage characteristics. After our survey findings, we describe the qualitative themes which emerged through our interviews.

4.1. Participant characteristics

Table 1 shows demographic information for the eleven robotic arm owners including their age, gender, diagnosis, and CUE-Q score. The median age of the eleven robotic arm owners was 35 (IQR 29.5–56.5). Among them, eight were male, one was female, and two identified as nonbinary. Additionally, all participants identified as Caucasian, with two also identifying as Hispanic. The CUE-Q upper extremity score is shown in column five indicating a user’s perceived difficulty in reaching, pushing, pulling, and grasping with a highest score of 128 indicating no difficulty.

Table 1.

Demographic information and capabilities of upper extremity function score (n = 11).

ID	Age	Gender	Diagnosis	CUE-Q total (max 128)	Skill	Arm use (years)	Jaco control	Jaco mode switch operation
P1	57	F	Spinal Muscular Atrophy (SMA) Type2	9	Int.	1.2	Joystick	Finger
P2	27	M	SMA Type 3	17	Expert	0.3	Joystick	Finger
P3	33	M	Spinal Cord Injury (SCI)	3	Int.	1.1	Joystick	Finger
P4	56	M	SCI	2	Int.	8	Joystick	Finger
P5	60	M	Muscular Dystrophy	36	Expert	8.4	Penta-switch	Finger
P6	30	M	Muscular Dystrophy	0	Expert	5	Mouth joystick	Head
P7	29	NB	Muscular Dystrophy	17	Int.	3	Mini joystick	Knuckle
P8	58	M	Multiple Sclerosis (MS)	3	Int.	3.5	Joystick	Head
P9	24	M	Cerebral Palsy	24	Int.	3.5	Head array	Head
P10	37	NB	Arthrogryposis	42	Int.	1.3	Joystick	Finger
P11	35	M	Amyotrophic Lateral Sclerosis (ALS)	1	Int.	0.5	Foot joystick	Foot

NB: Nonbinary; Int.: Intermediate robot arm operational skill.

Column six lists the owner’s self-described robotic arm skill level (novice, intermediate, or expert). Eight owners identified as Intermediate, and three as Expert. None of the participants self-identified as a Novice. The next column provides robotic arm ownership in years where the median duration owners had the robotic arm was 3 (IQR 1.2–3.9).

Columns eight and nine list the modality used for controlling the robotic arm, and how they switched between modes. We noted that all participants used the same control for their robotic arm as their wheelchair, with the exception of one who controlled their wheelchair with a joystick and the robotic arm with a separate penta-switch. This participant had partial use of both hands, a feature not shared by all participants. Ten participants mounted the robotic arm to their wheelchair 24/7, and one had it mounted to their bed while awaiting the delivery of their new wheelchair.

4.1.1. Technology preferences and attitudes

The demographic surveys also inquired about users’ attitudes toward technology, as displayed in Figure 2. Owners ranked their attitudes toward technology on a scale from 1 to 7, answering: “How accurately does each of the following phrases describe you?” (1 is Not at all Accurate, 7 is Extremely Accurate). These results are displayed below in a violin plot, where increased width represents a higher percentage of responses. For example, nine users indicated that the statement, “I like the idea of using technology to reduce my independence on other people” was Extremely Accurate (7), while only two rated it as 6. The accuracy values had a larger spread in response to the statement “I generally wait to a adopt a new technology until all the bugs have been worked out”. The horizontal bar represents the median accuracy value for that statement.

Figure 2.

Owner’s attitudes toward technology.

4.2. Activity characteristics

For characterizing robotic arm tasks, the demographic survey also asked users to list three common tasks they use the robotic arm for, shown in Figure 3.

Figure 3.

Common tasks performed with robotic arm.

At the end of the survey fourteen pre-defined tasks were enumerated each with a Likert scale asking users to rank the difficulty of performing those tasks with the robotic arm. Figure 4 shows owners’ difficulty ratings for fourteen tasks on a scale from 1 (no difficulty) to 10 (most difficult). A selection of N/A indicates that the task is not applicable because they cannot do it at all with the robotic arm. Among the fourteen tasks, pushing a button, grasping a cup, and drinking from a cup were considered the easiest tasks (with the exception of one user who uses a mouth joystick). The most challenging task was opening a bottle, with almost half of the owners stating that this task was not possible. Scratching their back and skewering food were also noted as difficult tasks with some indicating that they were not possible with the robotic arm.

Figure 4.

Difficulty of performing each task with a robotic arm, ranked from 1 (least difficult) to 10 (most difficult), with 11 (N/A – not applicable).

4.3. Thematic findings

We extracted eight interview themes shown in Table 2.

Table 2.

Themes from interview.

Theme
Section “The robot arm creates independence and a sense of choice.”
Section “Owner prefers caregiver if perceived effort or task time is too high.”
Section “Too much mode switching between wheelchair and ARM.”
Section “Operation requires concentration and focus for precise and multi-step manipulation tasks.”
Section “Personality and caregiver relationship influence owner’s approach to using robot arm for new tasks.”
Section “Owners are comfortable using the robot arm in society.”
Section “Initial training is sufficient but desire community resources.”
Section “Owners suggest robot arm improvements.”

4.3.1. The robot arm creates independence and a sense of choice

The first theme describes the robotic arm as a core benefit, “I think more than anything, the robotic arm gives me hope about the future. It’s the one thing that gives more independence” (P11). This independence leads to providing more choice over user’s own function, “It’s been indescribable the difference … I feel more independent than I have ever before, even though my condition is progressive … I don’t think I’ve ever been able to fully trust my body to do what needs to be done, so it has created choice in how I operate and live my daily life” (P2). Hydration is one of the most commonly performed independent tasks, one user said “it’s really great to be able to drink at will, pour a drink at will, without involving anybody else.” (P5).

User P1 shared several examples of how gaining independence helped reduce stress. They recounted their first action with the robotic arm: opening a drawer and disposing of an over-stretched garment their caregiver insisted on keeping because it was easiest to put on. In the second example, they described use of a Cough Assist machine which previously was “very claustrophobic to have somebody else’s hand pressing the mask to my face,” and now “for the first time, when I put it in the robot hand and I hold it to my own face, I’m not so scared.” Lastly, they described comforting themselves, “One of the biggest struggles I had was when my mom died all I wanted was to go under the covers and cry and stay for a day. I didn’t get that liberty because the caretaker comes … there’s no privacy and it’s kind of humiliating when you have to go to someone and say, ‘can you wipe my eyes’, and you don’t want to explain things … one of the greatest moments for me was when I realized that I could pull out my own tissue” (P1).

The robotic arm also offers users control over their activities. P11’s caregiver said, “if the robotic arm is holding the feeding tube (by P11), that gives me a free hand to do the suction rather than having to use both hands” (P11). This control also increases safety when navigating environments, “I take public transportation everywhere I go. I used to have to have help … pressing the elevator buttons, and if there was no station agent, then I would just kind of be screwed. Asking strangers for help … it’s awkward in the best-case scenario and dangerous in the worst-case scenario” (P7). The robotic arm also provides more control over personal effects, “if I wanted to buy something and I wasn’t with an attendant, I would have the person working there go into my backpack to get my wallet. Now I have a money pouch with Velcro on the back of it … on the robot hand, and then I extend it to the cashier … that’s an example where it is more time-consuming, but I personally feel more comfortable having more control over my items and not needing to have a stranger go into my personal belongings” (P10).

Lastly, this sense of choice reduced the mental pre-planning that was previously required before an attendant left for the day. Now, there are “many tiny tasks throughout the day that I can do without needing to plan ahead or specifically ask someone to do it for me” (P4).

4.3.2. Owner prefers caregiver if perceived effort or task time is too high

The second theme describes how users decide to use the robotic arm based on perceived effort, time, or complexity, often only opting for caregivers when tasks are too demanding. Users weigh the effort required for robotic arm tasks against the ease of asking for help. Participant P10 likened this to how energy is spent when they explained, “I kind of think of it as two different ways of using your time, or two different ways of expending energy. It does take a degree of mental energy to go up to a stranger and say, ‘Hey, I want to buy this box of cereal on the shelf, can you please put it in my basket?’ There’s a level of engagement that sometimes you would do. You do or don’t want to do that. And then there’s a level of concentration to do the arm” (P10).

Two users gave examples of how time was a factor, “if it’s a relatively easy task and it’s important to me, like staying hydrated all day, I figured out a way to do that with a little bit of assistance from my caregiver in terms of setup. … I don’t ask a caregiver to do anything because I like the caregiver doing it better. I only ask the caregiver to do things because it’s so much quicker than Jaco” (P8). Another user said, “It always comes down to speed and just efficiency, and whatever is going to be quicker; like eating dinner with Jaco is just an example of a quote unquote difficult task. It may take half an hour, forty five minutes, which is why I’d prefer somebody else just do it for me” (P2). This factor includes time pressures like work schedules, “I’m happy to do something myself, but if I’m under pressure because of my conference call for work during the day, I’ll probably ask my caretaker” (P1).

Users also gave some examples where their personal approach to task completion was the driving decision for doing tasks independently, “Nobody knows where stuff goes other than you. You’re the person that knows how you use your stuff and trying to explain to others … it’s sometimes easier to do it yourself” (P3).

Finally examples were also given based on caregiver availability, “If they’re doing dishes or something, I’d rather do it myself then wait for them to be done, but if they’re there and around and able to help and it’s convenient; then I would prefer them to help” (P7). One participant said, “when people aren’t around is when I use it the most” (P6).

4.3.3. Too much mode switching between robot arm and wheelchair

The next two themes emphasize the mental and physical demands placed on users, especially for complex manipulation tasks. Previous work has discussed how switching between the robot arm’s operation modes impacts task completion times [22]. However, our interviews revealed that the more frustrating mode switching occurs between the user’s wheelchair and the robotic arm where owners must cycle wheelchair modes before accessing arm operation. This operation causes a delay, “just think about going through life and everything you want to do with your arm has a 10 s delay” (P8). User P8 further illustrated this by describing opening a refrigerator, “If my wheelchair wasn’t in just the right position so that I could reach the item in the refrigerator, then I’ve got to cycle my wheelchair into drive and get closer and then see if that’s close enough … then cycle my wheelchair into Jaco mode, reach, and, oh darn, I didn’t get close enough, so I’ll make another adjustment.” This switching between the robotic arm and wheelchair is a common frustration for nine of the eleven owners. One user described this frustration while entering elevators, “I’ve had elevators where I press the button and it’s just a crappy elevator that would close immediately when it opens, so I’ve had issues where I’ve had to do acrobatics to get in the elevator, which is frustrating. If I was able to move my wheelchair while moving the arm, it would make the elevator easier.” (P7). However, two users with specialized setups (a penta-switch, and transition via long press) did not experience this issue.

4.3.4. Operation requires concentration and focus for precise and multi-step manipulation tasks

Eight of the eleven owners mentioned that concentration was required for arm operation. For example, higher multi-step tasks require more focus. One owner described refrigerator opening, “it’s no longer difficult, but it does require a lot of concentration … there’s a lot of potential for error, which could damage the robot … I need to get Jaco’s fingers in behind that handle to pull, and if I am negligent … you can easily break a Jaco finger” (P2). Eating was a second example, “You’ve got to concentrate. You’ve got to be paying attention. You’ve got to be thinking about how you are going to approach this item of food … you can’t necessarily just spear something with a fork; you’ve got to maybe come at an angle, or else it’s going to slide off that fork. So it’s an involved and very active thinking process while eating” (P2).

This concentration requirement includes instances where users multi-task, “Not more than a couple of weeks and a couple of cups of hot coffee in my lap, but where I miscounted (modes), and I still make a point to try not to be mid-conversation when trying to do something” (P5). One user almost dropped a cup when demonstrating their approach, “And now I will set the cup - whoa - I thought I was in elevate, but I was not, and I’ll blame that on talking … I’m not a good multitasker” (P8). Then another gave an example of being interrupted, “It takes a lot of concentration to use the arm and to pick up something very precisely. I’ve had situations when I’m out in public and I’m trying to do something kind of tricky and someone stops and is like, ‘Hey, can you explain to me how this works, this is really cool.’ I’m like, sure, I don’t mind telling you about it, but at the same time, I really can’t do the thing I’m trying to do because it’s so hard to multitask” (P10).

4.3.5. Personality and caregiver relationship influence owner’s approach to using the robotic arm for new tasks

Owners discussed their internal motivation for attempting new tasks with the robotic arm, revealing that this is influenced by both their personal attitude toward trying new things and their relationship with their caregiver. Personality traits include factors such as personal enjoyment, willingness to experiment, and considerations of mental energy in their interactions with the technology. One user described their stubborn approach, “If I’m by myself, I’m sometimes nonsensical and stubborn. There will be situations where, to save time and energy, I should just wait for somebody … but I’m stubborn, and I know I can get it. So, I will spend a half-hour trying to get it because I want to get it” (P7).

Owners who are more complacent and have family members providing 24/7 care discussed trying less with the robotic arm. For example, “I’ve got such good caregiving that it’s maybe a little bit spoiled that way that I don’t take Jaco to the point that other people do. … I’d much prefer the robot do it, but not at any cost. By cost, I mean, how complicated it is to achieve with a robot versus how easy it is to achieve with a human. We haven’t gone nearly as far as we could because my wife is retired and she’s right there. If I were to try to get a food item out of the refrigerator, bring it over and consume it myself, that would be an enormous feat for me and Jaco to do, but it would be just for the purpose of proving we could do it. It wouldn’t make my life easier or make my wife’s life easier” (P8).

Some owners’ caregivers encourage them to learn new tasks, “But I think they get excited when I find something I can do independently; they also feel excited for me … I think just that’s the way our relationships are sometimes we collaboratively figure things out … It’s been exciting for all of us to work with it and just see how it can make things a little easier and more independent for me” (P10).

4.3.6. Owners are comfortable using the robot arm in society

Owners express “if the arm benefits me that’s all that matters” (P9), and do not care about others’ perceptions. Owners also gave examples where the robotic arm is an interesting talking point in the community, “It’s gotten me a few free dinners in restaurants” (P5). One owner also felt the robotic arm helped challenge stereotypes about disability, “When you have a robotic device that’s obviously cool and advanced, and people see you using that, I think people start to think, that person’s life isn’t as bad as I thought it might be. Look at the interesting things that person can do. I don’t have to look at them as a different, you know, category of person, a person to be pitied, a sad person, just another person who is more relatable. I find that having something like the robotic arm frankly makes me a more interesting person and gives us things to talk about, and I really enjoy sharing technology with people and I don’t feel conspicuous” (P8).

However, one person also remarked how others’ fascination with the arm can be distracting. “I think one of the things I didn’t really anticipate is just how it changes the way that people interact with me. When I’m out doing things, in my head, I thought, oh, I’ll go to the store, get what I need, leave … People are curious about it. They stop, and they want to ask me about it, which sometimes is okay, but sometimes I’m in a rush, and I need to get to work or need to pick up something quickly” (P10).

4.3.7. Initial training is sufficient but desire community based resources

The training process was a positive experience for all owners. Once the robotic arm was acquired, a representative installed the arm, and provided guidance on basic operation. Owners said they only needed time to practice on their own, not more time with a technical representative. One user explained, “I would say after the initial demo session, I felt like I had a pretty good idea of how to do general things. And I mean, the first day I used it, I got it. … it probably took about a month or so before I could do things more efficiently, but pretty much from day one, I felt comfortable with all the basic maneuvers” (P7). They expressed the sentiment of learning by doing where owners felt they had mastered the basic controls within a couple of weeks to a couple of months.

Despite mastering basic maneuvers, owners mentioned wanting additional community resources to enhance these basic skills. For example one owner said, “I’m kind of in a funk. I’ve learned these four or five tasks, and I’m not really learning more … I’ve become complacent. The things I can control are all doable, but I may run out of ideas on how to accomplish a task or the task may just be so variable that it’s difficult” (P8). They suggested adding more training or online resources to provide new ideas. For example, one user suggested “An area to show you the technical parts and then an area to show you the practicality too, be nice to have both” (P11), and “a video for the caregivers … It would be nice to have the caregivers also be trained on the arm” (P11).

One owner suggested having accessories to enhance arm operation, where in addition to the arm, Kinova can “give you something to start with, like a plate, a cup that’s not going to spill or break so that you could try it for a while. I don’t know; it seems like they could have their own sort of online store. ‘Here are some things to go with Jaco’ ” (P1).

Owners also discussed a more comprehensive delivery program that builds on initial training by incorporating accountability and follow-up, such as inquiring about current arm usage and ways to expand it. Another suggested skillset improvement, “I think it would have been nice if there was an occupational therapist that was trained with the robot to do a little bit of training with it … I have one guy who takes care of the arm. I don’t think he’s the guy that’s going to sit there and work with you on what plate or what cup works for you or how you can adapt your toothbrush. I’m figuring that stuff out on my own” (P1).

4.3.8. Owners suggest robot arm improvements

During the interview, owners suggested robot arm improvements. These improvements are summarized in Figure 5. The first concern was a desire for better control over top speed settings (n = 5). While a Kinova representative mentioned users can adjust the speed, our interviews revealed a lack of awareness on how to easily make these adjustments. Another common request was for the ability to continue robot arm usage even after transferring to their bed (n = 3). With no current seamless transition between the wheelchair and bed, most users keep the arm always attached to their wheelchair.

Figure 5.

Owner quotes that describe the six areas most often cited as necessary for improving the robotic arm.

Users also expressed a desire for smoother, more fluid movement, noting frustration with the mechanical feel of frequent mode switching (n = 3). This desire for a less rigid operation is closely linked to concerns about the inconvenience of switching modes. Additionally, eight out of eleven users reported issues with the fragility of the gripper’s fingers, citing breakages and inadequate grip strength, pointing to the need for more durable components (n = 8).

Controlling the wrist, was described as the most challenging aspect of operating Jaco, but it was also seen as the most beneficial once mastered (n = 7). Users suggested that improving the intuitiveness of wrist control, especially when the reference frame changes, would enhance usability. Collision safety was another concern, with users recommending softer materials to prevent injury when the arm gets too close to their body or drops during power outages (n = 4).

5. Discussion by combining HAAT factors

By combining the analysis of the surveys and interviews, we gain valuable insights into the adoption and operation of the robotic arm. Using the HAAT model, we explore the relationships between power wheelchair users, their use of the robotic arm, and how they perform manipulation tasks in both home and community settings. We highlight survey findings and describe how the HAAT factors (human, activity, assistive technology, and context) shape usage through themes. Assistive technologies should not be evaluated solely on technical performance, but also on how the technology incorporates multiple outside factors to fit within a broader system of support for users’ daily activities.

5.1. Human perspective and factors

First, we describe a snapshot of the human or user where we presented demographic information for the eleven owners in Participant Characteristics section. Owners have an average CUE-Q score of 14 out of 128 suggesting high levels of upper limb functional limitations. They also self-identify as tech savvy exhibiting very positive perspectives toward technology, as shown in Figure 2. In terms of the owner’s robotic arm operating skill, none of the owners identified as novices. Therefore, the initial adoption of assistive robotic arms can be characterized by individuals interested in the benefits of technology with upper limb impairment that requires a high level of assistance.

Furthermore, personal characteristics such as risk tolerance, cognitive abilities, and motivation influence device usage.

Personal tasks like transferring, dressing, or bathroom activities not only require more physical strength and increased precision from the arm, but also the ability to maneuver the arm in tight spaces near the body, which users find risky. For example, toileting, a task that requires users to schedule their day around caregiver assistance, requires precise movements in limited spaces beyond a user’s field of view, which users avoid due to risk of discomfort and safety concerns. These difficulties highlight not only the physical limitations of the device but also the emotional hesitation users might have when operating it for personal tasks.

While basic operations of the robotic arm (e.g., using the joystick and switching between the four modes) are familiar to users, more complex tasks create mental strain. Controlling the wrist, especially at odd angles, and managing precise, multi-step activities like opening a door require high levels of focus (Theme 4.3.4). Users reported that long sequences of actions can be difficult, especially when multitasking or navigating social environments, such as using the robotic arm while conversing or dining out. This suggests that the mental demands of operating the arm, particularly when switching between the arm and wheelchair within the wheelchair menu (Theme 4.3.3), should be addressed. Improving the human-machine interface (HMI) to make complex operations more intuitive would be beneficial. As one user described, “If you think about your hands, your wrist is making adjustments to accommodate the flexibility of the door handle. The door is pretty stationary, but you have the extra axis of movement at your wrist. And for the robot hand, you have to make those movements separately. You have to move incrementally in straight lines” (P10).

In addition to task complexity, the psychosocial implications of using the robotic arm also influence task choices. For example, users may avoid tasks like eating with the robotic arm due to the time and effort required, preferring caregiver assistance instead. This is further complicated by personal motivation (Theme 4.3.5) since some users may attempt a task to demonstrate their capability but ultimately prefer assistance due to the social consequences (e.g., making a mess while eating), or perceived ease. Our work saw the choice of assistance between caregiver and Jaco was based on a user’s internal representation of effort in their current context (Theme 4.3.2), similarly recent work by [15] found this choice was associated with a user’s previous habit or perceptions on if operating the arm would be a burden. Understanding these factors requires recognizing that user decisions are shaped by both the technical limitations of the AT and their personal social and emotional context.

5.2. Activity factors

Devices that can assist with manipulation activities are important for those with upper limb impairment, as reach has been identified as one of the most desired maneuverability features according to a survey of 52 power wheelchair users [23]. Additionally, a study recording the life-log of a healthy individual’s daily activities found that 43% involved object lifting, with the majority of tasks requiring the use of hands and arms [24].

Owners listed common manipulation tasks they perform with the robotic arm as part of the demographic survey shown in Activity Characteristics section. These tasks are not all-encompassing as they only display statistics for tasks owners listed in the demographic survey. During the interviews, many owners also mentioned using the arm for daily tasks that arise as needed, such as flushing the toilet, closing bottle tops, or opening blinds. One owner with ALS mentioned that he uses the arm to wave because then he “feels more human” (P11). Additionally, three robotic arm users described how they use the arm to interact with their pets, “I wish you could see my service dog because she’s on the floor next to me lying on her back because she loves for me to open the claw to scratch her stomach with the arm” (P5). Lastly, owners also mentioned tasks they were most proud of doing at least once which included: carving a pumpkin, drinking a beer, grasping a Dorito, taking a selfie, DJ’ing by taking song requests, throwing the first pitch at Dodger stadium, eating Thanksgiving dinner, and making a box of mac n cheese. These example tasks demonstrate how the robotic arm has become an integrated part of the owners’ daily lives.

Despite the benefits of the robotic arm as an assistive technology, users still found certain tasks difficult, suggesting room for improvement. Tasks involving multiple movements (e.g., opening a fridge, cabinet, or door), those requiring greater situational awareness (e.g., scratching their back or retrieving overhead objects), or better grip strength (e.g., opening a bottle) were found to be more challenging. In contrast, simpler tasks with shorter action sequences, like grasping a cup, drinking, or pushing a button, were performed more frequently, as shown in Figure 4.

Users’ choice of tasks often reflects a balancing act between what the robotic arm can technically do and the ease or efficiency of completing those tasks. Simpler activities are more frequently attempted, while complex or multi-step tasks like opening doors or managing fine manipulation happen less. Although the robotic arm can perform a wide range of activities, users’ preferences toward caregiver assistance over using the arm are nuanced and depend on their perceptions of the effort required in different contexts (Theme 4.3.2). This can be affected by daily time demands or the ease of their relationship with caregivers.

5.3. Assistive technology (AT) factors and improvement

User’s ability to perform tasks were influenced by the design and usability of the robotic arm. For example, the robotic arm’s physical limitations were frequently mentioned, including issues with strength, grip durability, and the ability to maneuver in tight spaces. Users mentioned tasks requiring the arm to operate close to itself were more difficult than those where it could extend. Users also reported difficulty with small buttons and adjusting fingers to grasp items from the floor, “so you have to close a little bit, raise a little bit, close a little bit … by the time you do it, you drop whatever it is you’re trying to pick up” (P5). Additionally, bi-manual tasks like stretching headphones open or unscrewing bottle lids were not feasible.

Another challenge lies in the interface for controlling the robotic arm. While many users quickly grasped basic operations, they often hit a plateau in advancing their skills. They attributed this not only to the arm’s limitations in complex tasks but also a lack of continued training and resources (Theme 4.3.7). This plateau points to opportunities for more intuitive software control, autonomous support for repetitive tasks, and better long-term training. Future training should take a holistic approach, providing resources and motivation to help users continually improve their skillset through more comprehensive service delivery.

Owners also identified several key areas for improving the robotic arm, including speed control, bed attachment, smoother movement, better grip strength, more intuitive wrist control, and collision safety as shown in Figure 5. These user insights informed our recommended design enhancements outlined in Table 3. However, even with these adjustments, there may be limits to what the human-technology interface can achieve, highlighting the potential need for intelligent software to further enhance the robotic arm’s capabilities.

Table 3.

Design recommendations for Jaco robot arm.

Design Factor	User Suggestions	Design Recommendation
Modular attachment	Attach to bed. Attach to other ambulatory assistive devices. Less effort unclipping.	Modular design. Plug-n-play capability. Wheelchair detached power options.
Adjustable speed and grip strength	Adjust speed of ARM during operation. Improve grip strength. Less fragile fingers.	Finer more adaptive speed control. Improved gripper design.
Smoother operation	Desire smoother more fluid motion. Less mode switching for long-horizon repeatable tasks. Less mode switching between wheelchair and ARM.	Semi-autonomous functional improvement. Learned user profile for mode switching. Hot switch between wheelchair and ARM. Options for separate ARM control inputs.
Intuitive wrist movement	Intuitive control of reference frame switching. Intuitive wrist control.	Semi-autonomous functional improvement. Increase situational feedback to user.
Collision safety	Alternatives for controlled fall during device shutdown. Less hard edges.	Softer materials. More compliance near user.

5.4. Contextual factors: environment and social dynamics

In the home, the robotic arm provides users with a sense of independence (Theme 4.3.1), reducing their reliance on caregivers for tasks like grabbing a drink or moving objects. This autonomy instills a sense of hope,control, and choice over daily activities. The robotic arm provides significant benefits expanding users’ worlds, literally improving their ability to navigate environments. However, in public settings, while users feel comfortable operating the arm, it can sometimes draw positive or negative attention, which influences how efficiently users execute tasks (Theme 4.3.6). Additionally, practical considerations such as getting a wheelchair fully under a table or the robotic arm obstructing the line of sight during meals create situational challenges in the physical environment that limit its effectiveness in certain social settings.

Overall, usage of the robotic arm in social and outside settings was overwhelmingly positive (Theme 4.3.6), which contradicts the stereotype that some people with disabilities feel reluctant being seen with an AT device that draws unwanted attention to their disability [25]. The comfort experienced by robotic arm users can be attributed to several factors, including the advanced design and functionality of the device, which is often perceived as innovative and empowering rather than as a marker of dependency. Additionally, societal attitudes toward robotic technology may be more accepting, especially in a world where robotics is increasingly viewed as a tool of empowerment. In contrast, individuals using more traditional forms of assistive technology may face stigmatization due to lingering societal perceptions that equate such devices with physical limitations.

A significant factor influencing the use of the robotic arm is the relationship between the user and their caregiver (Theme 4.3.5). Many users see themselves and their caregivers as a collaborative unit when deciding how much effort to invest in a task. This balance between personal autonomy and caregiver reliance reflects the importance in considering both the user’s internal motivation and the broader social dynamics of care when assessing the impact of the robotic arm’s usage. Our finding reveal that the caregiver could be a facilitator or an unintentional barrier in regards to the user’s adoption or frequent use of technology which is a less explored research topic.

5.5. Owner advice to novice ARM users

Long-term owners benefit from their learned experience recognizing initial training was not a key factor, but continued skill development should be prioritized. Drawing from these experiences, we asked the users to share tips and advice to novice users, as peer-to-peer training is also strongly encouraged in the assistive technology field. Owners’ suggestions included, “Be patient,” “You never know until you try,” “It takes practice,” and “Challenge yourself to try different things with Jaco. They may not always work, but that’s Ok, learning comes from the process.”

A common tip from multiple owners was to always perform the finger-closing movement before anything else. This ensures gripped objects are not prematurely dropped when proceeding with a task. Many also emphasized the importance of mastering wrist modes, while some shared a trick of memorizing modes to the point where they no longer need to look at the mode screen.

5.6. Limitations

We recognize that our sample population is predominantly male, broader insights would be provided by including more female owners in future studies. Additionally, a significant number of participants identified as tech-savvy, potentially biasing their preferences for operating the robotic arm. Consequently, caution is advised in interpreting the results, as early adopters may possess a distinct inclination toward technology, differing from the perspectives of future users with varying technology experience and attitudes.

Our thematic analysis was also limited in having only one coder do final identification of remaining interviews. While we rigorously generated a Codebook through multiple meetings, which included coverage analysis, coding interviews five through eleven were by only one coder which might create bias. The authors minimized this bias by continuing to discuss the analysis in lab meetings to reach team consensus until theme saturation was achieved.

6. Conclusions and future work

Our study interviews eleven robotic arm owners across the United States and Canada, gathering feedback from users with ownership ranging from eight years to a few months. This paper examines both the benefits and challenges associated with the technical operation of robotic arms, as well as the importance of understanding users’ technical readiness and needs, which influence the adoption of these devices. By identifying sources of operational frustration and recognizing how intrinsic motivation can be fostered beyond technical improvements, we aim to enhance the performance and usability of robotic arms. Additionally, we provide guidelines for technology improvements, while recognizing the need to address service-related issues, such as training and caregiver dynamics, to further improve the user’s experience with owning and operating a robotic arm.

Future developments should prioritize intuitive controls, offer advanced user training, and improve adaptability to social and environmental contexts. Additionally, fostering collaboration between users and caregivers, and incorporating psychosocial factors into the design process, can further enhance the user experience.

Therefore, an additional contribution of this work is to highlight factors that influence the adoption of commercially available robotic arms, providing clinicians with insights to better assess and prioritize factors to those that may not be early adopters. For example, successful robot arm owners tend to be technically savvy, intrinsically motivated, and not discouraged by the cognitive demands required to operate the arm through multi-step tasks. They also benefit from caregiver support, aligning with previous studies that discuss the importance of the user [26] and caregiver involvement [15] in the service delivery process. Future users that do not meet this criteria benefit from additional training, support services, fostering supportive caregiver relationships, and resources to expand ideas on operational techniques. Effective assistive technology service delivery requires a holistic approach that addresses not just the technical operation of devices but also the broader context of user needs and caregiver dynamics.

➤ IMPLICATIONS FOR REHABILITATION.

• Gaining insight into the benefits and concerns of current Jaco robotic arm owners can inform future research, design, and the acceptance of wheelchair-mounted assistive robotic arms.

• Robotic arm owners require support and follow-up from practitioners, caregivers, and peers to enhance their robotic arm operation skills and increase usage.

• Broader contextual, human, and activity-related factors influence the effective use and adoption of assistive robotic arms.

Appendix A: Interview questions

We asked the following list of Interview questions:

• How has the robotic arm benefited you, and conversely has it created any barriers?

  • Has it increased your participation in activities?
  • Do you feel comfortable using the arm around friends, family, work, and in the community?
• How is your caregiver involved in the setup and support of the robotic arm?

  • Do they assist with the three robotic arm tasks you mentioned in the survey?
  • How much do you rely on your caregiver when controlling the robotic arm with your current control?
  • What tasks do you prefer your caregiver performs that does not involve the robotic arm?

• Do you use any environment adaptations in conjunction with the robotic arm?

• If your robotic arm is available, can you demonstrate using it to pick up an item nearby?

  • Can you state each action you make, such as changing control modes? Pretend you are describing your actions to someone who has never used the robotic arm.
• Based on your survey answers can you provide more details about the tasks that you listed which were easy to do with the robotic arm?

  • Is there a particular task that you were most proud to achieve using the robotic arm?
  • What is the most complex task you have done successfully with the robotic arm? Unsuccessfully?
  • Typical task time for completing “easy” tasks?
• Based on your survey answers can you provide more details about the tasks that you listed which were difficult to do with the robotic arm?

  • What actions are more difficult for you than others?
  • Is this a physical limitation with you controlling the arm or limitation of the robotic arm’s abilities?
  • Typical task time for completing “difficult” tasks?
• In the survey, you indicated three tasks you use the robot arm for. What task have you tried that you will never do again?

  • Are there any tasks you currently cannot do that you wish you could?
• Have you ever used preset button positions for the arm to move to? (Like the home button)

  • Have you ever used drinking mode and did you find it useful?
• Can you discuss how the robotic arm was delivered?

  • How long did it take to setup, and were you given any training at delivery time and/or setup time?
  • Were simple control features explained?
  • Were advanced control features explained?
• Have you received any formal training using the robotic arm? (Paid professional? Someone sent to your house?)

  • Have you received any informal training using the robotic arm? (Online Videos, self-taught)
  • How could the training process be improved?
• You self-described your robotic arm ability in the survey (novice/intermediate/expert). What robotic arm operating advice do you wish you could have known earlier?

  • Advice that you would give, knowing what you know now.
  • Do you have any tips or tricks for other robotic arm users?

The remaining interview questions about control modality preferences, along with a demonstrated video, were previously published in the paper by Wang et al. [27].