Portable Moxibustion Robot With Intelligent Acupoint Positioning Guided by Visual Sensing.

BACKGROUND

To overcome key challenges in traditional moxibustion, such as the shortage of skilled practitioners, time-consuming and labour-intensive procedures, this study developed an intelligent robotic system for automated acupoint positioning and moxibustion execution.

METHODS

The system established a standardised acupoint coordinate map utilising DensePose-based body normalisation and statistical modelling from 50 subjects. A hybrid serial-parallel robot design enabled large workspace and high precision for bedside application. Safety was ensured by a Model Predictive Controller utilising real-time thermal feedback for dynamic distance adjustment.

RESULTS

Experimental validation demonstrated precise spatial positioning, with mean errors of 5.8 ± 2.9 $5.8\pm 2.9$ mm (back), 6.4 ± 3.7 $6.4\pm 3.7$ mm (abdomen), 4.3 ± 2.7 $4.3\pm 2.7$ mm (upper limbs), and 9.4 ± 6.2 $9.4\pm 6.2$ mm (lower limbs). The system responded to temperature changes within 0.25 s, limiting overshoot to below 0.8°C.

CONCLUSIONS

This integrated robotic system provides a precise and scalable solution for automated moxibustion, particularly valuable for resource-limited clinical environments.

TRIAL REGISTRATION: It was an investigational clinical study, not a registrational clinical trial.