Control Architecture for Human-Like Motion With Applications to Articulated Soft Robots

Area of Science:

• Robotics
• Biomimetics
• Control Theory

Background:

• Human motor performance surpasses current robotic capabilities due to musculoskeletal properties and central nervous system control.
• Soft articulated robots mimic vertebrate musculoskeletal systems, but require advanced control for optimal function.
• Bridging the gap between biological and robotic systems necessitates replicating human-like motor control mechanisms.

Purpose of the Study:

• To introduce a control framework for articulated soft robots that ensures natural movements.
• To implement human central nervous system functionalities like learning, anticipation, and reactive replanning.
• To validate the proposed control architecture through simulations and experiments on a bio-mimetic robot.

Main Methods:

• A hierarchical control architecture with two levels: low-level dynamic inversion for trajectory tracking and high-level redundancy management.
• Implementation of functionalities such as learning by repetition, after-effect, anticipatory behavior, reactive re-planning, and state covariation.
• Validation using simulations and experimental testing on a bio-mimetic articulated soft robot.

Main Results:

• The proposed control framework successfully enables natural movements in articulated soft robots.
• The system demonstrated effective trajectory tracking, redundancy management, and precise task execution.
• Validation confirmed the framework's ability to replicate key human motor control functionalities.

Conclusions:

• The novel control framework effectively enhances motor performance in soft articulated robots.
• The approach successfully integrates human-like control strategies, advancing the field of bio-mimetic robotics.
• This research provides a robust foundation for developing more sophisticated and adaptable soft robotic systems.