Data-driven Shape Sensing of Continuum Dexterous Manipulators Using Embedded Capacitive Sensor

Area of Science:

• Robotics
• Sensor Technology
• Embedded Systems

Background:

• Continuum Dexterous Manipulators (CDMs) require accurate shape sensing for effective control.
• Existing Fiber Bragg Grating (FBG) sensors are costly and sensitive to temperature variations.
• There is a need for low-cost, robust shape sensing solutions for CDMs.

Purpose of the Study:

• To propose and evaluate a novel, inexpensive embedded capacitive sensor (ECS) for shape sensing in CDMs.
• To address limitations of current FBG sensors, focusing on cost and temperature sensitivity.
• To demonstrate the feasibility of ECS for real-time shape feedback and control.

Main Methods:

• Development of an inexpensive embedded capacitive sensor (ECS) prototype.
• Calibration of ECS using a vision-based system and a recurrent neural network (RNN).
• Evaluation on a 3D-printed, cable-driven CDM prototype with multiple markers.

Main Results:

• The ECS system achieved a 6.6% tip position error, normalized to the CDM's length.
• The calibrated ECS data accurately reflected the manipulator's shape.
• Successful computation of tip angle and position using data from three ECSs.

Conclusions:

• Embedded capacitive sensors (ECS) show early feasibility for shape sensing in CDMs.
• ECS offers a promising, cost-effective alternative to FBG sensors.
• Further improvements can enhance ECS performance for advanced CDM applications.