Hybrid parameter identification of a multi-modal underwater soft robot

Area of Science:

• Robotics
• Biomimetics
• Soft Materials

Background:

• Soft-bodied robots offer unique advantages in locomotion and manipulation.
• Octopus-inspired designs leverage natural systems for complex movements.
• Characterizing the dynamics of highly flexible, multi-modal robots is challenging.

Purpose of the Study:

• To introduce and characterize an octopus-inspired soft robot capable of both swimming and crawling.
• To develop and apply novel dynamic characterization techniques for soft, multi-modal robots.
• To evaluate the effectiveness of compartmentalized parameter identification for dual locomotion strategies.

Main Methods:

• A hybrid optimization approach combining least squares and genetic algorithms for parameter identification.
• Segregated characterization of swimming (pulsed-jet propulsion) and crawling (legged-locomotion) phases.
• Utilizing rubber-like materials for up to 80% of the robot's volume to enable flexibility.

Main Results:

• Compartmentalized parameter identification proved effective for characterizing the robot's dual ambulatory strategies.
• The study identified an underestimation of the quadratic drag coefficient when using static thrust recordings.
• Demonstrated the viability of ad hoc dynamic characterization techniques for soft-bodied, multi-modal robots.

Conclusions:

• Hybrid optimization and segregated parameter identification are viable protocols for multi-modal soft robot characterization.
• Static thrust recordings can lead to significant underestimation of drag coefficients in dynamic models.
• Further research is needed to refine dynamic modeling for shape-changing, self-propelled soft robots.