Modelling and implementation of soft bio-mimetic turtle using echo state network and soft pneumatic actuators.

Area of Science:

• Robotics and Biomimetics
• Soft Actuators
• Bio-inspired Design

Background:

• Mimicking biological movement in robotics is challenging due to complex morphology and flexibility.
• Soft pneumatic actuators (SPAs) offer potential for replicating animal-like motion.
• Turtle locomotion presents a specific bio-inspired design challenge.

Purpose of the Study:

• To investigate the use of SPAs to mimic turtle flipper motion.
• To design and simulate a bio-mimetic turtle flipper.
• To model the complex motion dynamics of a soft robotic turtle.

Main Methods:

• Finite element analysis (FEA) was used to simulate SPA behavior for 22 geometric configurations.
• Simulations were conducted across a pressure range of 0.11–0.4 MPa.
• An Echo State Network (ESN), a reservoir computing technique, was employed to model turtle kinematics.

Main Results:

• Optimized SPA designs were selected based on FEA simulations and validated with vision feedback.
• A bio-mimetic turtle was constructed using four SPAs with varying inclination angles.
• The ESN accurately modeled the turtle's motion path with a maximum root-mean-square error, correlating path with actuator rotation.

Conclusions:

• Soft pneumatic actuators can effectively mimic turtle flipper motion.
• Echo State Networks provide a viable method for modeling the nonlinear dynamics of soft robotic systems.
• This research advances the development of robots capable of flexible, naturalistic movement for improved interaction with living creatures.