A Smart Capacitive Sensor Skin with Embedded Data Quality Indication for Enhanced Safety in Human-Robot Interaction.

Area of Science:

• Robotics and Automation
• Sensor Technology
• Machine Learning

Background:

• Smart sensors are crucial for safety in human-robot interaction, especially with advancing machine learning in constrained environments.
• Monitoring data quality in real-time is essential as data-driven approaches are increasingly deployed on sensors.
• Existing systems require enhanced safety measures for reliable human-robot collaboration.

Purpose of the Study:

• To develop and evaluate a smart capacitive sensor system with integrated real-time data quality monitoring.
• To improve the safety and reliability of human-robot interaction applications.
• To enhance object detection accuracy through robust data quality assessment.

Main Methods:

• Implementation of a smart capacitive skin sensor using consumer-grade electronics for distance and angle detection.
• Integration of a dedicated software layer for real-time data quality monitoring.
• Utilizing a fully connected neural network for object position/angle inference and a one-class SVM for out-of-distribution data quality assessment.

Main Results:

• The sensor accurately detects object distance and angle within a 200 mm range under normal conditions.
• The embedded algorithm successfully identifies abnormal operating conditions related to poor data quality.
• Mean absolute distance error improved from 11.6 mm to 7.5 mm when data quality monitoring was active.

Conclusions:

• The developed smart capacitive sensor system with embedded data quality monitoring enhances human-robot interaction safety.
• Real-time data quality assessment is critical for optimizing sensor performance and reliability.
• This approach offers a significant advancement in creating safer collaborative robotic systems.