The Impact of Gait Pattern Personalization on the Perception of Rigid Robotic Guidance: A Pilot User Experience Evaluation

arXiv:2512.17425 · cs.RO · Submitted 2025-12-19 · Read on arXiv

Listen

Radio episode about this paper

Transcript

Introduction to the show: ident: Robotics Radio. Generated commentary on the latest robotics and control papers.

Rosa: I'm Rosa.

Dev: And I'm Dev, here with Taro.

Rosa: We're looking at the paper: "The Impact of Gait Pattern Personalization on the Perception of Rigid Robotic Guidance".

Dev: Here's the gist of the paper. Exoskeletons modulate human movement across diverse applications from performance augmentation to daily-life assistance.

Rosa: Subjective evaluations revealed minimal differences between gait pattern conditions.

Dev: Pruning redundant reasoning tokens The objective of the study was to determine the impact of gait pattern personalization on user perception of rigid robotic guidance.

Rosa: Personalization was achieved using a data-driven framework that predicts hip, knee, and pelvis trajectories from walking speed, anthropometric, and demographic data.

Paper summary: Rosa: So this paper we're looking at is called "The Impact of Gait Pattern Personalization on the Perception of Rigid Robotic Guidance: A Pilot User Experience Evaluation". It’s about exoskeletons that move people, like for helping them walk better or just to keep them moving when they need to. The main thing they're trying to figure out is whether making the walking pattern specific to an individual actually changes how comfortable or natural the experience feels for the user.

Dev: Right, so it’s looking at whether tailoring the gait pattern improves perception of rigid robotic guidance, which matters because these systems are often enforced by design. The core idea seems to be that there's little understanding about what users actually *think* when they walk with these imposed patterns compared to a standard one.

Taro: I wonder if this is just looking at the setup, or if they’re really testing how it plays out when things go wrong in real life? Because exoskeletons are rigid, so any mismatch between the pattern and what the person *wants* could be a big issue for safety or comfort.

Rosa: Exactly. The paper says they tested ten participants in a treadmill-based exoskeleton and used three different gait patterns: personalized, standard, and a randomly selected one from their database. They are seeing if these subjective experiences—things like comfort and enjoyment—actually differ between the patterns they tried out.

Dev: And what they found is that overall comfort actually improved significantly when comparing the third trial to the first one, which suggests adaptation to the setup plays a role, even with personalization. But for things like interest or passiveness metrics, there weren't any significant differences found among those conditions.

Taro: So if you look at it from a failure mode perspective, does this mean that when the robot enforces a standard pattern versus a personalized one, the way participants put force into those knee actuators changes? Like what are the mechanical implications of that force difference?

Rosa: Well, they did some interaction forces analysis on the mean absolute force at both knee actuators. They found that when people were forced to follow a random gait pattern compared to the standard ones, they reported significantly higher force when following that random pattern. But comparing personalized and standard patterns only showed a significant difference on the left leg.

Dev: That’s an interesting detail because it suggests the personalization effect isn't as uniformly strong as some might expect across both legs or conditions. It also brings up a point about the control systems, since they used a rigid position-derivative controller, which was chosen to make sure they could actually measure these forces well.

Paper summary: Taro: And what about the limitations here? Since this was a pilot study with ten people, they’re admitting that the dataset they used for personalization didn't fully represent all the ways people move in real life. They also mentioned that the rigid position-derivative controller might have actually contributed to some of that discomfort participants felt.

Rosa: So, looking at what this paper says about its limitations, it seems like they’re flagging a couple of things: the dataset wasn't big enough to capture all the movement variability, and they are worried that the way the controller was built might be causing issues for users. They suggest future work should look at different cuff designs and more compliant control strategies.

Dev: That leads into why this matters for people using these devices. If you only listen to this show, it means that while tailoring a walking pattern can make someone feel better in the short term because they adapt to the system, it doesn't necessarily translate into a fundamental change in how they perceive the guidance itself.

Taro: So for someone who is just listening, what does this actually mean for their day-to-day use of an exoskeleton? Does it mean I should focus more on how well my body adapts to the robot, or is personalization something that really matters long term?

Rosa: The conclusion they draw from this pilot study is that participants generally rated the gait pattern they experienced last as being more comfortable and natural compared to the first one, which they suggest is just adaptation to the experimental setup. So, it points toward personalization having a reduced short-term impact compared to adapting to how the entire robotic system feels ergonomically.

Dev: It sounds like a lot of this research is focused on user experience metrics like comfort and enjoyment, rather than just pure kinematic precision or control loop performance. The implication is that for real-world application, designing the system around overall ergonomics might be more important than getting every single gait parameter perfectly individualized right away.

Taro: I think that's where the autonomy research comes in—how does this subjective feeling translate into a reliable safety case when things get messy outside the lab? If a personalized pattern feels good now but breaks down when you hit an unexpected obstacle, that’s a different kind of failure mode we need to address.

Rosa: Exactly. So, the paper on "The Impact of Gait Pattern Personalization on the Perception of Rigid Robotic Guidance: A Pilot User Experience Evaluation" shows us that while we can personalize the walking patterns using data about speed and size, the actual subjective improvement in comfort is mostly temporary adaptation.

Dev: It highlights that we need to keep exploring how these users interact with rigid guidance systems, because even if personalization doesn't drastically change their perception of naturalness right away, it’s still a design choice worth making before we build something for everyone.

Conclusion: Rosa: So we're wrapping up this look at how tailoring walking patterns affects what people feel when they use these rigid exoskeletons.

Dev: It boils down to this pilot study, "The Impact of Gait Pattern Personalization on the Perception of Rigid Robotic Guidance," and it was done by a few researchers.

Rosa: The main thing they found is that while you can personalize the walking pattern based on someone's height and speed, the actual comfort level didn't change dramatically for most people.

Dev: They tested personalized, standard, and random patterns on ten participants. And yeah, overall comfort actually went up when they moved from the first trial to the third one because people adapted to being in the setup more.

Rosa: But when it came to interest or how passive they felt about the guidance, those metrics didn't show any significant differences between the patterns.

Dev: That’s interesting because it suggests that for this type of rigid assistance, adapting your body to how you're wearing the device matters more than having a perfectly personalized walking path in terms of enjoyment.

Rosa: They also looked at the forces—how much muscle they had to use at the knees—and they saw some differences there, but only on one leg when comparing personalized versus standard patterns.

Dev: That points to where we need to be careful; if you're designing a control loop, you can't just focus on one side of the body for force application.

Rosa: The authors themselves noted that their data wasn't big enough to cover all the different ways people move in real life, so they aren't claiming this is a universal solution yet.

Dev: Exactly. It means we need to keep designing these systems with overall ergonomics in mind rather than just chasing every single personalized kinematic detail right away.

Rosa: So what does this mean for us when we think about putting these devices into everyday use?

Cognitive Robotics Department, Delft University of Technology · Department of Rehabilitation Medicine, Erasmus Medical Centre, Rotterdam, The Netherlands · Department of Biomechanical Engineering, Delft University of Technology, Delft, the Netherlands · Institute of Automatic Control, RWTH Aachen University · Department of Mechanical Engineering, FUM Center of Advanced Rehabilitation and Robotics Research (FUM CARE), Ferdowsi University of Mashhad, Mashhad · Walker Department of Mechanical Engineering, University of Texas at Austin

cs.RO

Submitted: 2025-12-19

Updated: 2026-04-10

DOI: 10.3389/frobt.2026.1937649

License: http://creativecommons.org/licenses/by/4.0/

Importance score: 85/100

The gist: Exoskeletons modulate human movement across diverse applications from performance augmentation to daily-life assistance.

Key concepts

Gait Pattern Personalization
This involved creating custom walking patterns for each user based on their unique physical data like height and speed. The system predicted how a person's hip, knee, and pelvis should move while walking to create a tailored experience.
Standard Gait Pattern
This was the baseline or average walking pattern derived from the entire database of volunteers. It served as the control condition against which personalized and random gait patterns were compared during testing.
Rigid Robotic Guidance
This refers to a robotic system that enforces specific, predetermined movements on a person. The study examined how different ways of guiding movement—personalized versus standard—changed the user's feeling of discomfort or enjoyment.
Adaptation to System Ergonomics
The findings suggest that users' comfort improved more from getting used to the robotic equipment and its overall design than from simply using a specific personalized walking pattern. This means the physical setup itself had a stronger effect on perception.

Terminology

Summary

Exoskeletons modulate human movement across diverse applications from performance augmentation to daily-life assistance. p. 0

Where the paper says it

1 The Impact of Gait Pattern Personalization on the Perception of Rigid Robotic Guidance: A Pilot User Experience Evaluation Beatrice Luciani1 , Katherine Lin Poggensee1,2 , Heike Vallery3,4 , Alex van den Berg1 , Severin David Woernle1 , Mostafa Mogharabi1,5 , Stefano Dalla Gasperina1,6 , and Laura Marchal-Crespo1,2 Abstract— Objective: Exoskeletons modulate human movement across diverse applications, from performance augmentation to daily-life assistance. These systems often enforce specific kinematic patterns to mitigate injury risks and motivate users to keep moving despite diminished capacity.

Subjective evaluations revealed minimal differences between gait pattern conditions. p. 8
Where the paper says it

We developed and evaluated a comprehensive kinematic model for the exoskeleton control and regressionbased predictive models generating personalized hip, knee, and pelvis trajectories from anthropometric, demographic, and walking speed data. Subjective evaluations revealed minimal differences between gait pattern conditions.

Pruning redundant reasoning tokens

The objective of the study was to determine the impact of gait pattern personalization on user perception of rigid robotic guidance. Personalization was achieved using a data-driven framework that predicts hip, knee, and pelvis trajectories from walking speed, anthropometric, and demographic data. p. 0

Where the paper says it

Methods: Ten unimpaired participants completed a within-subject experiment in a multi-planar treadmill-based exoskeleton that enforced three different gait patterns: personalized, standard, and a randomly selected pattern from a publicly available database. Personalization was achieved using a data-driven framework that predicts hip, knee, and pelvis trajectories from walking speed, anthropometric, and demographic data.

The standard pattern was obtained by averaging gait patterns from the aforementioned database. p. 0
Where the paper says it

Personalization was achieved using a data-driven framework that predicts hip, knee, and pelvis trajectories from walking speed, anthropometric, and demographic data. The standard pattern was obtained by averaging gait patterns from the aforementioned database.

How the training signal is built

The gait pattern prediction models were developed to predict leg joint and pelvis trajectories based on individual characteristics such as height and age. The approach builds on the models proposed by Koopman et al. p. 2

Where the paper says it

Gait pattern prediction models We developed models to predict leg joint and pelvis trajectories based on individual characteristics such as height and age. Our approach builds on the models proposed by Koopman et al.

, which use multiple polynomial regression models to predict knee and hip flexion/extension joint trajectories from a person’s height and walking speed. p. 2
Where the paper says it

Our approach builds on the models proposed by Koopman et al. [31], which use multiple polynomial regression models to predict knee and hip flexion/extension joint trajectories from a person’s height and walking speed.

To improve the personalization potential, other factors like age, body weight, and gender were also accounted for. p. 2
Where the paper says it

To improve the personalization potential of Koopman’s proposed models, here we also account for the effect of other factors (e. , age, body weight, and gender), which have been shown to substantially affect joint angles and overall walking patterns [11], [12], [32], [33].

The database used for training included data from 42 volunteers across different age groups and walking speeds. p. 2
Where the paper says it

This database includes data from 42 volunteers, split into 24 young adults (21 years to 37 years) and 18 older adults (50 years to 84 years), all without lower-extremity injuries or gait impairments. Data were collected with a markerbased motion-capture system composed of twelve cameras, from participants walking barefoot on a treadmill at eight different speeds, ranging from 40 % to 145 % of their comfortable, self-selected walking speed.

Where the gains come from

The user experience analysis showed that overall comfort significantly improved in the third trial with respect to the first one. No significant differences were perceived among conditions regarding Interest/Enjoyment and Passiveness metrics. p. 5

Where the paper says it

Questions related to the Smoothness of movements and whether the limbs were pushed beyond their natural range did not show significant differences across conditions or condition order. The models for both the Interest/Enjoyment (IMI) and Passiveness metrics revealed no significant differences, either in terms of gait pattern or condition order.

The Personalized pattern got the highest enjoyment score, but differences with the other patterns were not significant. p. 5
Where the paper says it

The models for both the Interest/Enjoyment (IMI) and Passiveness metrics revealed no significant differences, either in terms of gait pattern or condition order. The Personalized pattern got the highest enjoyment score, but differences with the other patterns were not significant (β = 0.

Interaction forces analysis

The results for the LMM analysis of the mean absolute force at both knee actuators reported a significantly higher force when participants were enforced to follow the Random gait pattern compared to Standard gait patterns. The difference between Personalized and Standard gait patterns only approached significance on the left leg. p. 6

Where the paper says it

002; Left leg: β = 84. The difference between the Personalized and Standard gait patterns only approached significance on the left leg (Right leg: β = 22.

Study limitations

Limitations included the dataset not fully representing anthropometric variability and the small sample size. The rigid position-derivative controller, selected to amplify perceptibility, might have contributed to participants’ discomfort. p. 7

Where the paper says it

Uncontrolled factors, such as personal footwear— standardized in the reference dataset [34]—may also have influenced perception. Importantly, the rigid position-derivative controller, selected intentionally to amplify the perceptibility of the trajectories, might itself have contributed to participants’ discomfort.

Future work should explore alternative cuff designs and compliant control strategies. p. 8
Where the paper says it

The little differences observed in our study, therefore, emphasize the need to explore whether more permissive, trajectory-free, or compliant controllers would allow personalized kinematic features to manifest more clearly in subjective experience and objective outcomes [25]. Future work should systematically investigate the aforementioned factors by testing alternative cuff designs and interface stiffness, comparing fully guided and assistive/impedancebased control modes, and implementing coupled prediction models that explicitly encode inter-joint dependencies.

Conclusions

In this within-subject user experience pilot study, we explored user perceptions of personalized versus standard and random gait patterns using an exoskeleton capable of multiplanar motion assistance. We found that participants generally rated the gait pattern experienced last as more comfortable and natural compared to the first one, suggesting adaptation to the experimental setup. p. 8

Where the paper says it

Subjective evaluations revealed minimal differences between gait pattern conditions. We found, however, that participants generally rated the gait pattern experienced last as more comfortable and natural compared to the first one, regardless of the gait pattern enforced, suggesting adaptation to the experimental setup.

These findings indicate that personalization of gait kinematics has a reduced short-term impact compared to adaptation to the robotic system and overall system ergonomics. p. 8
Where the paper says it

We found, however, that participants generally rated the gait pattern experienced last as more comfortable and natural compared to the first one, regardless of the gait pattern enforced, suggesting adaptation to the experimental setup. These findings indicate that, in a small representation of unimpaired users, personalization of gait kinematics has a reduced short-term impact compared to adaptation to the robotic system and overall system ergonomics.

Green — the paper says it, and the page is printed. Red — no passage carries it, so treat that sentence as unverified. The same check runs on your own text: check a summary against its source, $5.

Improvements for AI systems

  1. Adaptive User Experience Modeling: AI systems can incorporate a time-dependent adaptation factor into their control loops, recognizing that the third trial was generally perceived as more comfortable (p = 0.037) and more natural (p = 0.037) compared to the first trial. This allows the system to dynamically adjust its guidance strategy or stiffness based on the user's current level of adaptation, rather than relying solely on a fixed personalized trajectory.

  2. Inter-joint Coupling Prediction: AI models should be developed to explicitly encode inter-joint dependencies when generating trajectories, as the study notes that personalization predicts the different joints’ trajectories separately, without taking into consideration their inter-joint coupling. This would improve trajectory generation by ensuring that predicted movements are kinematically consistent across connected body segments.

  3. Compliance and Control Strategy Integration: The AI should explore using compliant control strategies to mitigate discomfort, as rigid guidance was noted to reduce compliance at the joints and lead to a general sensation of unnaturalness. This could involve implementing energy-shaping controllers or impedance-based control modes that allow for natural variability around the desired trajectory.

  4. Preference-Informed Refinement: Future AI systems should incorporate a feedback loop where personalized gait adjustments are refined based on user input, as the paper suggests that trajectory generation should not only aim for accurate prediction but also allow refinement (e.g., step length or width adjustments) based on individual user feedback.

  5. Robustness to Kinematic Variability: AI models should be trained to account for the inherent variability of healthy human gait, recognizing that one person’s average gait trajectory is not rigidly distinct from another’s but is instead likely to fall within the broader distribution of kinematic patterns expressed by other healthy young adults. This would make personalized guidance more robust across a wider population.

Abstract

Exoskeletons modulate human movement across diverse applications, from performance augmentation to daily-life assistance. These systems often enforce specific kinematic patterns to mitigate injury risks and motivate users to keep moving despite diminished capacity. However, little is known about users' perception of such robot-imposed guidance, especially when personalized to the uniqueness of individual human walk. Given the usually substantial computational cost for personalization, understanding its subjective impact is essential to justify its implementation over standard patterns. Ten unimpaired participants completed a within-subject experiment in a multi-planar treadmill-based exoskeleton that enforced three different gait patterns: personalized, standard, and a randomly selected pattern from a publicly available database. Personalization was achieved using a data-driven framework that predicts hip, knee, and pelvis trajectories from walking speed, anthropometric, and demographic data. The standard pattern was obtained by averaging gait patterns from the aforementioned database. After each condition, participants rated enjoyment, comfort, and perceived naturalness. Knee joint interaction forces were also recorded. Subjective ratings revealed no significant differences among patterns, despite all trajectories being executed with high accuracy. However, gait patterns experienced last were rated as significantly more comfortable and natural, indicating adaptation to the system. Higher interaction forces were observed only for the random vs. standard pattern. Personalizing gait kinematics had minimal short-term influence on user experience relative to the dominant effect of adaptation to the exoskeleton. These findings highlight the importance of integrating subjective feedback and accounting for user adaptation when designing personalized robot controllers.

Related papers