SpiKon-E: Hybrid Soft Artificial Muscle Control Using Hardware Spiking Neural Network.

Area of Science:

• Robotics
• Materials Science
• Biomimicry

Background:

• Artificial muscles are crucial for advanced robotics and medical devices, shifting from servo motors to smart materials.
• Current control strategies for artificial muscles require optimization to precisely mimic natural muscle actuation.

Purpose of the Study:

• To design and implement a novel artificial muscle system that replicates biological muscle's full range of motion and control.
• To advance sustainable soft robotics through innovative actuation and control mechanisms.

Main Methods:

• Developed a novel shape memory alloy (SMA)-based linear actuator with a guiding mechanism for increased displacement.
• Integrated the SMA actuator into a hybrid soft actuation structure with a silicone PneuNet and force sensor.
• Utilized a hardware Spiking Neural Network (HW-SNN) for real-time force control at the actuator's endpoint.

Main Results:

• The novel SMA linear actuator demonstrated significantly higher displacements compared to traditional SMA wire-driven systems.
• The integrated hybrid soft actuation system provided real-time pressure feedback and precise force control.
• Experimental results confirmed superior displacement and performance compared to traditional control-based SMA systems.

Conclusions:

• The developed artificial muscle system successfully mimics biological muscle actuation, offering enhanced displacement and control.
• This research contributes significantly to sustainable soft robotics, particularly in actuation and control for humanoid robotics.