⚡ electrical engineering

Locomotion Variability and User Experience in Smart Wheelchair Human-Robot Interaction

This paper demonstrates that an autonomy-supportive shared control strategy for smart wheelchairs, which preserves users' natural movement variability rather than suppressing it, significantly enhances perceived agency and usefulness without compromising task performance.

Original authors: Sean Kille, Adina M. Panchea, Balint Varga, Sören Hohmann

Published 2026-08-13

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Sean Kille, Adina M. Panchea, Balint Varga, Sören Hohmann

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

Imagine you are learning to ride a bike. At first, you wobble a lot, your handlebars twitch, and your path is a bit messy. But as you get better, you don't just become a perfect, straight-line robot. You learn to make tiny, natural adjustments to stay balanced. You might sway slightly left or right when the wind blows, or take a slightly different route around a puddle. This "wobble" isn't a mistake; it's actually a sign of a healthy, adaptable system. In the world of science, this is called variability. It's the idea that human movement is naturally messy and unpredictable, but that messiness follows a pattern: we are very steady when we need to be precise (like stopping at a red light) and more flexible when it doesn't matter (like cruising down a straight road).

Now, imagine a robot trying to help you. For a long time, engineers thought the best way to help was to act like a strict coach who says, "Stop wobbling! Be perfect!" They built robots that tried to smooth out every little shake and force the human to move in a perfectly straight, boring line. But this paper asks a big question: What if the robot is actually making things worse by trying to be too perfect? What if, by trying to eliminate all the natural wobbles, the robot is taking away our feeling of control? This research sits at the intersection of robotics and human psychology, exploring whether letting humans be a little messy actually makes them feel more in charge and happier with the help they receive.

The Smart Wheelchair Experiment

In this study, researchers set up a clever experiment using an Intelligent Powered Wheelchair (IPW). Think of this wheelchair not just as a chair, but as a robot that can "feel" how you push it and offer a little nudge to help you steer. The goal was simple: push the wheelchair from a starting point, through a doorway, to a finish line. But the researchers wanted to see how different types of "help" changed the experience.

They tested three different ways of driving:

The "No Help" Mode: You push the chair entirely on your own. This was the baseline to see how you naturally move.
The "Strict Coach" Mode (lowVar): The robot constantly corrects your path, smoothing out every little wobble to keep you on a perfect, straight line. It's like having a robot hand on the handlebars that never lets you drift even a millimeter.
The "Smart Buddy" Mode (highVar): This is the new, special mode the researchers invented. Here, the robot acts like a good coach. It lets you wiggle and sway freely when you are in the middle of the room (where it doesn't matter much), but it steps in to give you a firm, steady hand only when you are approaching the doorway or the finish line (where you need to be precise).

What They Found

The results were surprising and quite fun. When the researchers looked at the data, they saw that the "Smart Buddy" mode successfully kept the natural, bell-shaped pattern of human movement. Just like in the "No Help" mode, people were wobbly in the middle of the room and steady at the start and finish. The "Strict Coach" mode, however, flattened everything out, making the path look like a rigid ruler.

Here is the big takeaway: The "Smart Buddy" mode didn't make the task harder. In fact, people were just as accurate at reaching the finish line in the "Smart Buddy" mode as they were in the other modes. The robot didn't ruin the performance; it just let the human be human.

But the real magic happened in how the participants felt.

The Feeling of Control: When people used the "Strict Coach" mode, they reported feeling much less in charge. It felt like the robot was taking over, and they were just along for the ride. Their "sense of agency"—that feeling that I am the one doing the action—dropped significantly.
The Happy Factor: In contrast, the "Smart Buddy" mode made people feel much more in control, almost as much as when they had no help at all. They also rated this mode as more "useful" and "easy to use" compared to having no help at all.

One participant even described the difference perfectly: in the "Strict Coach" mode, the wheelchair felt like it was stuck on rails, forced to go exactly where the robot wanted. But in the "Smart Buddy" mode, the wheelchair felt lighter and easier to control, like the robot was a partner rather than a boss.

Why This Matters

The paper suggests that we don't need to treat human movement like a problem to be fixed. For a long time, engineers thought that variability (the wobbles) was just "noise" that needed to be cleaned up. This study suggests that noise is actually a feature, not a bug. By designing robots that respect our natural, messy movement patterns—helping us only when we really need it—we can build systems that make us feel more capable and autonomous.

The researchers are careful to say that while these results are promising, they are based on a specific test with a small group of people (18 participants). They found that preserving natural variability didn't hurt performance and definitely improved how people felt about the robot. It suggests that the future of helpful robots shouldn't be about forcing humans to be perfect machines, but about building partners that understand and respect our natural, slightly wobbly, human rhythm.

Technical Summary: Locomotion Variability and User Experience in Smart Wheelchair Human–Robot Interaction

Problem Statement
Human movement is inherently variable and stochastic, yet model-based Human–Robot Interaction (HRI) assistance strategies often assume deterministic behavior. By treating human variability as noise to be minimized or suppressed, conventional control systems may inadvertently alter the user's experience, potentially reducing their sense of agency. While the functional benefits of variability in motor learning are recognized, its deliberate preservation in shared control settings—and its specific impact on user experience in physically coupled systems—remains underexplored. This study addresses the gap in understanding how different assistance strategies that either suppress or preserve natural locomotion variability affect both objective task performance and subjective user experience.

Methodology
The authors conducted a user study involving an Intelligent Powered Wheelchair (IPW) operating as a shared control system. The system utilized a differential drive platform with force sensors on the handles to capture human input. The study compared three distinct automation modes:

No Support (noSup): A baseline condition with no automation assistance, used to capture each participant's natural movement patterns and variability.
Low Variability Assistance (lowVar): A conventional assistance mode providing continuous rotational guidance along a mean path, effectively suppressing variability throughout the trajectory.
High Variability Assistance (highVar): A novel, autonomy-supportive mode that selectively preserves natural movement variability. This strategy uses a "variability factor" ( $\lambda(k)$ ) modeled as an inverted bell-shaped curve. It applies maximum assistance (high gain) in task-relevant regions (start and goal points) to ensure precision, while minimizing assistance (low gain) in task-irrelevant regions (the mid-section of the path) to allow natural human variability.

Experimental Design
Twenty-one participants performed a point-to-point navigation task, pushing the IPW from a start position through a doorway to a goal. Each participant completed 15 repetitions under each of the three conditions. The order of the assisted conditions was pseudo-randomized.

Objective Measures: Task-irrelevant position variability (maximum variance of path error), endpoint error, endpoint variability, and human work (cumulative rotational input).
Subjective Measures: Sense of Agency (SoA), Perceived Usefulness and Ease of Use (via QUEAD2), and overall User Experience (via UEQ).

Key Contributions

Novel Control Framework: The design of a shared control framework that explicitly incorporates human natural variability patterns. It selectively allows variability in task-irrelevant areas while maintaining assistance in task-relevant regions, contrasting with traditional approaches that uniformly reduce variability.
Empirical Evaluation: A systematic user study providing evidence on the effects of preserving versus restricting variability on both task performance and user experience in a physically coupled human-machine system.
Design Principles: Evidence that preserving natural variability does not compromise task-relevant performance and can significantly enhance the user's sense of agency and perceived usefulness.

Results

Objective Performance: The study confirmed that the highVar mode successfully preserved higher task-irrelevant variability compared to the lowVar mode, closely mimicking the natural bell-shaped variability pattern observed in the noSup condition. Crucially, there were no significant differences in task-relevant performance metrics (endpoint error and endpoint variance) between the three modes. This indicates that preserving variability did not degrade task success.
Subjective Experience:
- Sense of Agency: Participants reported a significantly higher sense of agency in the highVar mode compared to the lowVar mode. The highVar mode was comparable to the noSup baseline, whereas the lowVar mode significantly diminished the sense of agency.
- Perceived Usefulness: The highVar mode received significantly higher ratings for perceived usefulness compared to the noSup mode.
- Ease of Use: HighVar showed a marginally significant improvement in perceived ease of use compared to noSup.
- Overall User Experience: Both assisted modes (lowVar and highVar) received significantly higher user experience ratings than the noSup mode, though no significant difference was found between the two assisted modes.

Significance and Claims
The paper claims that variability-aware assistance is a viable design principle for human-centric HRI systems. The findings suggest that restricting a human's inherent variability negatively impacts the interaction experience, specifically by lowering the sense of agency. Conversely, assistance strategies that respect the embodied structure of human movement—by supporting task-relevant precision while allowing flexibility in task-irrelevant dimensions—can support both performance and user autonomy.

The authors modestly conclude that while variability-preserving designs show promise for balancing technical performance with positive user experience, the results are exploratory. They note that the use of personalized reference trajectories (derived from each user's natural path) may have attenuated the differences between the assisted modes, particularly regarding usability. The study emphasizes that treating variability as noise to be eliminated is less effective than designing systems that leverage the functional structure of human movement variability. Future work is suggested to explore personalized variability patterns, perceptual thresholds of variability, and the application of these principles to more complex, non-locomotion tasks.

The Smart Wheelchair Experiment

What They Found

Why This Matters

Technical Summary: Locomotion Variability and User Experience in Smart Wheelchair Human–Robot Interaction

More like this