Researchers have developed a multimodal interactive physiotherapy robot that combines multiple therapy techniques with high-precision dynamic acupoint recognition, addressing key limitations in existing systems. The robot integrates vision, touch, force, position, and language modalities to enhance safety and compliance during treatment.
Key takeaways
- RTOP-AcuDet model achieves an average error of 5.52 mm, recall of 91.43%, and detection speed of 31.5 frames/s for back acupoints.
- A pneumatic end-effector simulates various therapist hand movements, including pressing, vibrating, pushing, and kneading, with adjustable stiffness.
- A two-stage acupoint processing method and a standardized dataset of 200 volunteers support robust dynamic recognition across different body types.
- Force/position hybrid control and task planning enable accurate reproduction of physiotherapy techniques, with real-time user voice feedback for adjustments.
This system offers a promising solution for automated, personalized physiotherapy, potentially improving efficiency and accessibility in rehabilitation and chronic pain management for aging populations.
Source: 《机器人》期刊 (robot.sia.cn) · Published 2026-03-25 · “兼具多种理疗手法和穴位识别能力的多模态交互理疗机器人”