Haptic Suit Improves Motion Feedback with Simplified Design

A new haptic suit developed by researchers at Nagoya University can now provide more accurate motion feedback to users, addressing a long-standing in wearable technology. The device, called Funabot-Upper, uses artificial muscles to simulate 14 different upper-body movements, including those of the trunk, shoulder, elbow, and wrist. This advancement is significant because it improves upon previous designs that often caused users to confuse sensations between different body parts, such as mixing up shoulder and elbow motions. By refining how the suit stimulates the body, the researchers have created a tool that could enhance virtual reality experiences, rehabilitation training, and other applications where precise motion feedback is crucial.

The key finding from this study is that separating joint stimulation from muscle stimulation dramatically reduces perceptual mixing, where users previously struggled to distinguish between motions like shoulder adduction and elbow flexion. In experiments, the new suit achieved a recognition accuracy of 94.6% for seven specific motions when users were given feedback, up from 68.8% with the previous design. This improvement came from a simplified design policy that uses fewer artificial muscles—42 compared to 80 in the earlier version—while still covering more motions. The researchers confirmed that joint stimuli, which apply force across joints, are perceived correctly about 90% of the time when focusing on body part identification, whereas muscle stimuli, which pull on muscles, are less distinct but still contribute to the overall sensation.

Ology involved developing Funabot-Upper based on a new design policy that categorizes stimuli into two types: joint stimuli (Type J) and muscle stimuli (Type M). Type J stimuli are positioned across human joints to apply force along the motion direction, while Type M stimuli are placed along muscles to apply shear or compression force. The suit uses thin McKibben artificial muscles arranged in elliptical shapes on elastic polyester underwear and gloves, as shown in Figure 1 of the paper. These muscles are controlled by a pneumatic system that adjusts pressure to contract and relax them, with response times measured at 0.896 seconds for contraction and 1.442 seconds for relaxation. The researchers conducted experiments with eight male participants in their 20s, who wore the suit and reported perceived motions without prior training, using a setup where they stood and looked at a questionnaire to avoid visual cues.

From the experiments, detailed in confusion matrices in Figures 7 and 8, show that Type J stimuli led to an average correct recognition rate of 73.8%, with specific motions like elbow flexion perceived 100% of the time. In contrast, Type M stimuli had a lower accuracy of 37.5%, often evoking a pulling sensation rather than a specific motion. The comparison experiment with the previous Funabot-Suit demonstrated that the new design improved accuracy from 68.8% to 94.6% under feedback conditions, with significant gains in shoulder and elbow regions where perceptual mixing was most problematic. The data also revealed that the suit uses 67.5% fewer artificial muscles than before, simplifying the system and potentially reducing costs and weight, as noted in the paper's discussion of system simplification.

Of this research extend to real-world applications where haptic feedback is essential, such as in virtual reality gaming, remote operation of robots, or physical therapy for patients recovering from injuries. By providing more intuitive and accurate motion sensations, Funabot-Upper could make these technologies more effective and accessible. For example, in rehabilitation, it could help patients relearn movements without confusion, while in entertainment, it could enhance immersion by delivering clearer feedback during interactive experiences. The study's on separating joint and muscle stimulation offer a design guideline for future haptic devices, potentially leading to full-body suits that mimic natural movement more closely.

Limitations of the study include the use of a small, homogeneous sample of eight male participants with similar body parameters, which may not represent broader populations. The paper notes that perception rates for some motions, like shoulder horizontal extension and elbow extension, were lower due to the standing position restricting natural movement ranges. Additionally, muscle stimuli were less effective than joint stimuli, indicating room for improvement in how these are designed and applied. Future work, as mentioned in the conclusion, will focus on optimizing pneumatic response times and enhancing muscle stimulus perception to make the suit even more versatile and natural-feeling for users.