Natural movement timing enhances ownership of robotic arms

Summary

When AI powered prosthetic arms that move autonomously become widespread, understanding how people feel about them and accept them will be crucial. In this study, we used virtual reality to simulate a situation in which a participant's own arm was replaced by a robotic prosthetic arm, and examined how the prosthesis movement speed affects embodiment, including body ownership, the sense of agency, usability, and social impressions of the robot such as competence and discomfort. We found that both overly fast and overly slow movements reduced body ownership and usability, whereas a moderate speed close to natural human reaching, with a movement duration of about one second, produced the most positive impressions.

Contents

When a person loses a hand or arm, prosthetic limbs are essential technologies for maintaining everyday function. To date, much prosthetics research has focused on control methods that enable the device to move according to the user's intention, often by using biosignals such as electromyography (EMG) and electroencephalography (EEG), and on improving the accuracy of such control. Meanwhile, advances in machine learning and AI are making it increasingly realistic that future prostheses will assess the situation and provide assistance through autonomous or semi-autonomous movements. However, when a body part moves independently of one's will, people are likely to experience it as "unsettling" or "not part of my body," creating a major barrier to acceptance.

Addressing this issue, prior work has suggested that even if a limb moves on its own, discomfort can be reduced and acceptance as part of the body can increase when the movement's goal or intention is understandable. Building on this idea, Harin Manujaya Hapuarachchi and his colleagues (Hapuarachchi was a doctoral student at the time of the study and is now an Assistant Professor in the School of Informatics at Kochi University of Technology) focused on movement speed. In virtual reality, we presented an avatar whose left forearm was replaced with a prosthetic limb, and participants performed a reaching task. The prosthetic arm (a virtual forearm) autonomously flexed toward a target, and we systematically varied its movement duration across six levels (125 ms to 4 s). After each condition, participants rated body ownership, sense of agency, usability (SUS), and social impressions of the robot (RoSAS: competence, warmth, and discomfort).

The results were clear.

• At a moderate speed (movement duration of 1 s), body ownership, agency, and usability were highest.

• In the fastest (125 ms) and slowest (4 s) conditions, body ownership, agency, and usability were significantly lower.

• Perceived competence was higher at moderate to slightly faster speeds, whereas discomfort was highest in the fastest condition. Warmth did not show a clear dependence on speed.

These findings indicate that, in a future where AI enabled prostheses provide autonomous assistance, it is not sufficient to pursue faster and more accurate performance alone. Instead, movement speed should be designed to match what people can readily accept as part of their own body.

The insights may inform not only the design of autonomous prosthetic arms, but also other forms of robotic body augmentation, such as supernumerary robotic limbs, exoskeletons, and wearable robots, that operate as functional extensions of the body.

Looking ahead, we will also examine adaptation and learning through long-term use. People can come to experience familiar tools as if they were part of their body. If a fast and accurate robotic body part is used continuously in daily life, it may become "normal," feel easier to use, and be more readily embodied.

Finally, using VR is important because it allows researchers to safely simulate prosthetic technologies and control schemes that are not yet widely available, enabling the psychological, acceptance-related, and design requirements to be evaluated in advance.