An Enhanced Robot Massage System in Smart Homes Using Force Sensing and a Dynamic Movement Primitive.

Area of Science:

• Robotics
• Human-Robot Interaction
• Artificial Intelligence

Background:

• Smart homes and healthcare are increasingly integrating service robots to enhance quality of life.
• Robots with human behavioral sensing are crucial for personal use, reducing labor costs and human fatigue.
• Developing intuitive teaching interfaces for robots performing complex tasks like massage is an ongoing challenge.

Purpose of the Study:

• To propose a novel force-sensing and robotic learning-based teaching interface for robot-assisted massage.
• To enable robots to learn and generalize human-like movements from physical demonstrations.
• To ensure safe and accurate human-robot interaction during the massage task.

Main Methods:

• Human operators demonstrate massage movements by guiding the robot's end-effector.
• Finite Difference (FD) method for data segmentation, Dynamic Movement Primitive (DMP) for movement modeling, and Dynamic Time Warping (DTW) for multi-demonstration preprocessing.
• Gaussian Mixture Model (GMM) for DMP evaluation, Gaussian Mixture Regression (GMR) for trajectory synthesis, and a hybrid position/force controller for precise tracking and safety.

Main Results:

• The proposed method successfully models and generalizes human-like massage movements from demonstrations.
• The system synthesizes accurate trajectories, minimizing position errors through GMR.
• Validation on a KUKA LBR iiwa robot platform demonstrated effective and safe execution of massage tasks.

Conclusions:

• The developed force-sensing and robotic learning interface provides an effective method for teaching robots complex physical tasks like massage.
• The integration of DMP, GMM, and GMR enables robots to learn from human demonstrations and perform human-like movements.
• The hybrid controller ensures safe and precise human-robot interaction, paving the way for advanced service robots in homes and healthcare.